[{"content":"The confidence paradox Mid-summer thought: the more I learn, the less convinced I am that certainty is something intelligent people eventually reach. It seems to work the other way around. Every answer exposes another assumption, every assumption opens another question, and every question reveals a larger territory that I had not even realised was there. That is what keeps pulling me back to the Dunning–Kruger effect and not the internet-meme version where stupid people are confident and clever people are tortured by self-doubt, because the science is far more nuanced than that, but the much more interesting human problem underneath it: we are remarkably bad at knowing how much we know. I keep wondering whether that blind spot is really a defect at all. Evolution did not need us to possess perfect self-awareness; it needed us to survive, act, reproduce, build alliances and occasionally do something wildly ambitious before we had enough information to prove it was sensible. A species that understood every risk with perfect clarity might still be sitting in a cave discussing the downside scenarios of fire. Some degree of unjustified confidence may therefore have been incredibly useful. You need it to lead, to fight, to start a company, to fall in love, to invent something, or simply to say, “I think we should go this way,” when nobody can possibly know the outcome. Human progress probably owes quite a lot to people who were more certain than the evidence justified. The problem is that self-awareness seems to pull us in the opposite direction: the more seriously we examine what we know, the more obvious it becomes that there is always another layer. Are we therefore coded for infinite curiosity rather than final understanding? Is there even an endgame to intelligence, or is consciousness simply a machine that turns answers into better questions forever?\nWhen confidence scales That creates a strange conflict with leadership, because the world rarely rewards uncertainty as enthusiastically as it rewards confidence. The person who says “I need to think about this” can sound hesitant; the person who says “obviously, this is the answer” can sound decisive. We say we want thoughtful leaders, but we often promote the ones who project certainty most convincingly. Perhaps that was manageable when a confidently wrong person could only mislead a village, a boardroom or, on a particularly productive day, an entire country. Now confidence scales globally in seconds. Social media compresses complexity into declarations, algorithms reward outrage more reliably than nuance, and expertise often suffers from the unfortunate disadvantage of knowing why the simple answer is probably incomplete. I see the same thing in technology all the time: the person who has just discovered a platform knows exactly how everything should work; the person who has implemented it repeatedly starts asking irritating questions about context, integrations, incentives, geography, regulation and all the other details that destroy a beautiful PowerPoint. Knowledge ruins simple answers. That does not mean we have evolved into idiocracy; I suspect we have simply built systems that allow idiocy to scale faster than wisdom. And now we are adding AI to the equation. In the best version of the future, AI becomes the most powerful mirror we have ever built: something that can challenge our assumptions, argue the opposite case, expose what we are missing and force us to confront the possibility that our favourite conclusion is wrong. In the worst version, it becomes the greatest confidence-laundering machine in human history. I ask a question about something I barely understand, receive a beautifully structured answer, and five minutes later I am explaining it to someone else with the confidence of a man who personally discovered gravity. AI can expand our capability dramatically, but it can also make the gap between what we can produce and what we genuinely understand almost invisible. That may not cure Dunning–Kruger. It may industrialise it.\nThe real endgame So perhaps the goal is not perfect self-awareness at all. Extreme self-awareness without the ability to act becomes paralysis, while confidence without self-awareness becomes delusion. The useful place is somewhere uncomfortably between the two: being able to make a decision with conviction while keeping a small door open in your mind marked, “I may be completely wrong.” That, to me, is a far more interesting model of leadership than pretending uncertainty has disappeared. It is also where I hope AI can make us better without making us less human. I do not want technology to remove our irrationality, emotion, instinct or curiosity; those are not bugs to be patched out of the species. I want it to help us see ourselves more clearly without stealing the courage that makes us move. Maybe our blind spots were useful because certainty gave us momentum. Maybe curiosity exists because certainty was never supposed to last. And maybe the next stage of intelligence, human and artificial together, is not about finally having all the answers, but becoming better at holding two thoughts at once: I might be wrong, and I am moving anyway.\nMaybe the wisdom is simply this: knowing enough to act, while remaining curious enough to know you could be wrong.\n","permalink":"https://borggrech.com/thoughts/confidence-was-a-feature/","summary":"\u003ch2 id=\"the-confidence-paradox\"\u003eThe confidence paradox\u003c/h2\u003e\n\u003cp\u003eMid-summer thought: \u003cstrong\u003ethe more I learn, the less convinced I am that certainty is something intelligent people eventually reach.\u003c/strong\u003e It seems to work the other way around. Every answer exposes another assumption, every assumption opens another question, and every question reveals a larger territory that I had not even realised was there. That is what keeps pulling me back to the Dunning–Kruger effect and not the internet-meme version where stupid people are confident and clever people are tortured by self-doubt, because the science is far more nuanced than that, but the much more interesting human problem underneath it: \u003cstrong\u003ewe are remarkably bad at knowing how much we know.\u003c/strong\u003e I keep wondering whether that blind spot is really a defect at all. Evolution did not need us to possess perfect self-awareness; it needed us to survive, act, reproduce, build alliances and occasionally do something wildly ambitious before we had enough information to prove it was sensible. A species that understood every risk with perfect clarity might still be sitting in a cave discussing the downside scenarios of fire. Some degree of unjustified confidence may therefore have been incredibly useful. You need it to lead, to fight, to start a company, to fall in love, to invent something, or simply to say, “I think we should go this way,” when nobody can possibly know the outcome. Human progress probably owes quite a lot to people who were more certain than the evidence justified. The problem is that self-awareness seems to pull us in the opposite direction: the more seriously we examine what we know, the more obvious it becomes that there is always another layer. \u003cstrong\u003eAre we therefore coded for infinite curiosity rather than final understanding?\u003c/strong\u003e Is there even an endgame to intelligence, or is consciousness simply a machine that turns answers into better questions forever?\u003c/p\u003e","title":"Confidence Was a Feature. AI May Turn It Into a Bug."},{"content":"There is a strange feeling you get in Malta sometimes. You can be stuck in traffic, surrounded by cranes, apartments rising everywhere, restaurants full, delivery vans moving, and ships bringing in what the island cannot produce for itself.\nOn the surface, it looks like progress: more activity, more construction, more consumption, more money moving. But underneath it sits a harder question: how much more can a small place absorb before growth starts to feel less like progress and more like pressure?\nMalta is not the world, but it can feel like a preview. Small islands expose limits faster. Land becomes scarce, congestion becomes visible, and dependence on imports becomes impossible to ignore. For now, Malta can rely on the outside world. The planet cannot.\nThat is the uncomfortable reality. The economic story we inherited is built around expansion: more people, more customers, more transactions, more output. For a long time, that story worked. Growth created opportunity, reduced poverty, and expanded choice. But every system eventually meets its limits, and today\u0026rsquo;s limits are not only financial. They are physical, ecological, social, and psychological.\nThis matters because AI is about to put pressure on the old model from another direction. People still talk about AI as a productivity tool, but it is much bigger than that. It will reshape how work is done, how value is created, how expertise is priced, and eventually how societies organize themselves.\nA model built on people trading labour for money becomes unstable when machines can perform more of that labour. A model built on endless consumption becomes harder to justify when production becomes increasingly abundant. The challenge is not the technology itself; it is that our values were shaped by a world where scarcity was the default.\nWe still measure status through accumulation, treat busyness as proof of importance, and often confuse economic activity with human progress. But what if the next era requires a different instinct? Not more for the sake of more, but better. Cleaner. More intentional. More focused on quality of life than quantity of output.\nCapitalism is exceptionally good at turning desire into production. It is less good at deciding when enough is enough. AI may amplify that weakness before it helps solve it. If production, content, and services become cheaper, we may simply create and consume more of everything, including noise, distraction, and waste.\nThe danger is not that AI replaces jobs. The deeper risk is that it accelerates an old operating system before we have agreed on a new one. A society already obsessed with growth may use AI to grow faster. A market already optimized for attention may use AI to capture even more of it.\nWe tell ourselves that every problem created by growth can be solved by more growth. Sometimes that is true. Often it is just faith.\nThe next era may require a different measure of success: not how much we can produce, but how well we can live. Not how much value we can extract, but how much dignity, resilience, and wellbeing we can preserve.\nThis is not an argument against capitalism. It is a recognition that systems designed for one set of conditions can become dangerous when those conditions change. AI is not a small adjustment to the economy; it may change the logic of the economy itself. Currency, work, and even democracy will all feel that pressure.\nThe real risk is not that humanity fails to invent powerful tools. We are very good at that. The risk is that we keep using those tools to pursue outdated goals: more, faster, bigger, cheaper.\nAt some point, a mature civilization must ask a harder question:\nWhat should we stop optimizing for?\nThat may be the real reset, not the end of markets, ambition, or progress, but the end of pretending that growth alone is wisdom.\nA simple rule of thumb If a system can only survive by consuming more forever, it is not successful.\nIt is unfinished.\n","permalink":"https://borggrech.com/thoughts/abundance-will-break-the-old-growth-story/","summary":"\u003cp\u003eThere is a strange feeling you get in Malta sometimes. You can be stuck in traffic, surrounded by cranes, apartments rising everywhere, restaurants full, delivery vans moving, and ships bringing in what the island cannot produce for itself.\u003c/p\u003e\n\u003cp\u003eOn the surface, it looks like progress: more activity, more construction, more consumption, more money moving. But underneath it sits a harder question: how much more can a small place absorb before growth starts to feel less like progress and more like pressure?\u003c/p\u003e","title":"Abundance Will Break the Old Growth Story"},{"content":"Outline The tension: different people can ask AI similar questions and get different answers Why this happens: weak context, poor framing, and public models optimized to be helpful The flawed response: treating AI as proof instead of an instrument The better framing: AI should support judgment, not replace it What changes when you verify before believing Example: two people using AI to defend opposite positions Rule of thumb: never outsource judgment Open with the real problem AI is not dangerous because it is always wrong.\nIt is dangerous because it can make people feel completely right.\nI keep seeing the same pattern: someone asks an AI tool a question with weak context, receives a confident answer, and then treats the result as if they consulted the sum of human knowledge.\nThe problem is not only the answer.\nThe problem is the confidence it gives the person holding it.\nWhy this keeps happening Most people underestimate how much the quality of an AI answer depends on the quality of the question.\nContext matters.\nFraming matters.\nMissing facts matter.\nAssumptions matter.\nIf you ask a shallow question, you often get a shallow answer written in polished language. That polish creates an illusion of authority.\nAnd public AI models tend to be agreeable. They are designed to be helpful, cooperative, and responsive. That can be useful when you know what you’re doing.\nBut when the user is wrong, poorly informed, or emotionally invested in a conclusion, that agreeableness can reinforce the wrong belief.\nThe tool does not only answer the question.\nIt often reflects the shape of the question back to the user.\nThe tempting but wrong response The lazy response is to use AI as proof.\n“AI said I’m right.”\n“I checked it.”\n“That’s a hallucination.”\nThis is where the conversation becomes difficult.\nIt is hard to argue with someone who feels they have all human knowledge behind them, especially when pride enters the room.\nAnd there is a particularly frustrating move that appears more often now: when someone uses the tool incorrectly, gets a poor answer, and then dismisses disagreement by calling the other response a hallucination.\nSometimes hallucinations are real.\nBut sometimes “hallucination” is just a convenient label for information that challenges a weak position.\nThat is dangerous.\nBecause once people use AI to defend their pride, truth becomes secondary.\nA better way to think about it AI is not the final judge.\nIt is a thinking instrument.\nA very powerful one.\nBut still an instrument.\nIt can help structure arguments, surface patterns, pressure test ideas, explain unfamiliar concepts, and compare options. It can accelerate thinking.\nBut it cannot carry responsibility for your judgment.\nThat remains yours.\nThe mature posture is not blind trust or cynical rejection.\nIt is disciplined use.\nAsk better questions.\nProvide context.\nInterrogate assumptions.\nCheck sources.\nTest the opposite position.\nAsk where the answer may be incomplete.\nThe value of AI increases when the human becomes more rigorous, not less.\nWhat changes when you apply this You stop treating AI outputs as conclusions.\nYou treat them as drafts.\nHypotheses.\nMaps that may be useful, but may also be distorted.\nThat shift changes the entire interaction.\nInstead of asking, “What is the answer?” you ask:\nWhat assumptions is this answer making? What evidence supports it? What would a strong opposing view say? What facts would change the conclusion? Is this answer true, or merely plausible? This is where judgment lives.\nNot in receiving the answer.\nIn knowing how to test it.\nConcrete example Imagine two people debating a business decision.\nOne believes the company should automate a customer facing workflow immediately.\nThe other believes automation should be introduced more carefully, with review points and escalation paths.\nBoth ask AI.\nThe first person asks:\n“Why should we automate this process to increase efficiency?”\nThe AI gives a strong answer about productivity, speed, cost reduction, and scalability.\nThe second person asks:\n“What are the risks of automating this customer facing process without enough governance?”\nThe AI gives a strong answer about quality drift, trust loss, bad data, and unclear accountability.\nBoth people now feel validated.\nBoth can say, “AI agrees with me.”\nBut the tool did not resolve the question.\nIt amplified the framing.\nThe real work is not asking which answer sounds better.\nThe real work is defining the context:\nWhat is the workflow? What is the risk level? What happens if the output is wrong? Who owns the outcome? What review is required? What evidence do we have from similar cases? Only then does AI become useful.\nWithout that, it becomes a mirror with a confident voice.\nWhat I’ve learned the hard way AI rewards the person who can think clearly before using it.\nIt punishes the person who wants certainty without doing the work.\nThe tool is powerful, but it does not remove the need for discipline. In many cases, it increases it.\nBecause now anyone can generate a polished argument.\nThe question is whether the argument has a factual foundation.\nWhere to start When AI gives you an answer, do not stop there.\nStart your verification loop:\nCheck the factual claims against source material Look at links and primary references where possible Ask the model to argue the opposite position Ask another model the same question with the same context Speak to someone who understands the domain Speak to someone outside the domain who can expose hidden assumptions This may sound slower.\nIt is not.\nIt is faster than making a confident decision on weak foundations and paying for it later.\nA simple checklist Did I give the AI enough context to answer responsibly? Is the answer factual, or merely plausible? What assumptions is the answer making? Have I checked the sources behind the claims? Have I asked for the strongest opposing view? Have I asked someone with real domain knowledge? Am I using AI to think better, or to defend what I already wanted to believe? A rule of thumb Never outsource judgment.\nAI can help you think, but it cannot absolve you from being responsible for what you believe, repeat, or decide.\n","permalink":"https://borggrech.com/thoughts/ai-is-not-the-final-judge/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: different people can ask AI similar questions and get different answers\u003c/li\u003e\n\u003cli\u003eWhy this happens: weak context, poor framing, and public models optimized to be helpful\u003c/li\u003e\n\u003cli\u003eThe flawed response: treating AI as proof instead of an instrument\u003c/li\u003e\n\u003cli\u003eThe better framing: AI should support judgment, not replace it\u003c/li\u003e\n\u003cli\u003eWhat changes when you verify before believing\u003c/li\u003e\n\u003cli\u003eExample: two people using AI to defend opposite positions\u003c/li\u003e\n\u003cli\u003eRule of thumb: never outsource judgment\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eAI is not dangerous because it is always wrong.\u003c/p\u003e","title":"AI Is Not the Final Judge"},{"content":"Outline The tension: many of the strongest beliefs in society are rarely examined properly Why this keeps happening: assumptions are protected by habit, status, and social comfort The flawed response: either accepting inherited ideas blindly or attacking them carelessly The better framing: break beliefs down to fundamental truth and test whether they still serve reality What changes when you think across boundaries instead of inside them Example: the child who keeps asking “why?” until adults run out of borrowed answers Open with the real problem Some beliefs survive for a surprisingly long time without being tested.\nNot because they are true.\nBecause they are useful. Familiar. Socially protected.\nYou see it everywhere.\nIn work. In politics. In culture. In family life. In business. In education.\nA phrase gets repeated enough times and eventually it becomes untouchable.\nAt that point, questioning it is treated as a problem in itself.\nNot because the question is weak.\nBecause the belief has become part of the furniture.\nWhy this keeps happening Most people do not examine assumptions all the way down.\nThey inherit them.\nFrom parents. From school. From institutions. From the current mood of the culture.\nThat is understandable. No one has time to rebuild every belief from scratch.\nBut inherited thinking has a cost.\nA belief can make perfect sense in the world that produced it and become useless — or even harmful — in the world that follows.\nThe problem is that people often confuse age with truth, popularity with accuracy, and moral confidence with depth.\nSo assumptions survive long after their original conditions have disappeared.\nThe tempting but wrong response There are two lazy ways to deal with inherited beliefs.\nThe first is obedience.\nAccept the assumption because it is established, widely repeated, or emotionally protected.\nThe second is shallow rebellion.\nAttack the assumption because it is fashionable to be provocative, or because rejecting it signals independence.\nBoth fail for the same reason: neither is interested in truth.\nOne protects the idea. The other performs against it.\nNeither actually breaks it down.\nNeither asks the harder questions:\nWhy did this belief emerge?\nWhat need did it serve?\nWhat conditions made it sensible?\nAnd do those conditions still exist?\nA better way to think about it Attack assumptions carefully.\nEspecially the taboo ones.\nNot because taboo is exciting, but because protected beliefs often hide the most intellectual laziness.\nThe goal is not destruction.\nThe goal is essence.\nStrip away the language, the inherited emotion, the social packaging, and ask what remains.\nSometimes the belief still contains something durable.\nSometimes it is a fossil.\nThis kind of thinking gets stronger as your knowledge widens.\nBecause the deeper you go into multiple disciplines, the more you realize that the borders between them are artificial.\nEconomics touches psychology.\nPsychology touches biology.\nBiology touches energy.\nEnergy touches politics.\nPolitics touches incentives.\nIncentives shape culture.\nCulture shapes work.\nReality is one system.\nOnly human organization breaks it into departments.\nWhat changes when you apply this You stop thinking in isolated boxes.\nYou begin noticing patterns that repeat across domains.\nYou become less impressed by labels and more interested in mechanisms.\nYou ask:\nWhat is actually happening here?\nWhat force is really at work?\nWhat is the system optimizing for?\nWhich assumptions are left over from another era?\nThat shift produces better judgment.\nIt also produces better ideas.\nBecause many original ideas are not completely new.\nThey are truths transferred from one domain into another by someone who noticed the boundary was fake.\nConcrete example There is a reason children can be so unsettling.\nA child can sit in front of an adult and ask “why?” five times in a row.\nWhy do we do it that way?\nBecause that’s how it works.\nWhy does it work that way?\nBecause that’s the rule.\nWhy is that the rule?\nBecause that’s what people do.\nWhy?\nWhy?\nWhy?\nEventually the adult runs out of original thought and starts handing the child borrowed answers.\nThat moment is revealing.\nBecause many adult beliefs are exactly that: borrowed answers repeated with enough confidence to avoid further questioning.\nThe child is not smarter in a technical sense.\nBut the child is still close to first principles.\nMost adults are not.\nThey have learned the social skill of stopping their questions at the point where the group becomes uncomfortable.\nThat makes them easier to live with.\nIt does not make them more correct.\nWhat I’ve learned the hard way The more deeply you examine reality, the less respect you have for rigid intellectual borders.\nAnd the more carefully you question assumptions, the more you realize that many ideas survive on social pressure long after they have lost explanatory power.\nA rule of thumb If an idea cannot survive honest questioning, it does not deserve your loyalty.\nAnd if a boundary between disciplines prevents insight, the boundary is probably artificial.\n","permalink":"https://borggrech.com/thoughts/reality-does-not-respect-our-categories/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: many of the strongest beliefs in society are rarely examined properly\u003c/li\u003e\n\u003cli\u003eWhy this keeps happening: assumptions are protected by habit, status, and social comfort\u003c/li\u003e\n\u003cli\u003eThe flawed response: either accepting inherited ideas blindly or attacking them carelessly\u003c/li\u003e\n\u003cli\u003eThe better framing: break beliefs down to fundamental truth and test whether they still serve reality\u003c/li\u003e\n\u003cli\u003eWhat changes when you think across boundaries instead of inside them\u003c/li\u003e\n\u003cli\u003eExample: the child who keeps asking “why?” until adults run out of borrowed answers\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eSome beliefs survive for a surprisingly long time without being tested.\u003c/p\u003e","title":"Reality Does Not Respect Our Categories"},{"content":"Outline The tension: people claim they would happily work 24/7 if the salary were high enough Why this keeps appearing: social media reduces life decisions to financial trade-offs The flawed response: believing unlimited work is acceptable if the compensation is high enough The better framing: money cannot override biological limits, human needs, or lost life experience What changes when you treat time and life experience as the real scarce resources Example: the difference between temporary intensity and permanent sacrifice Open with the real problem You see this comment everywhere.\nSomeone shares a story about working extremely hard: long nights, relentless pressure, difficult trade-offs.\nThen someone replies:\n“If I was paid twice my salary, I’d happily do that.”\nAt first glance, it sounds practical. Even rational.\nBut look closer.\nThe comment assumes something impossible: that a higher salary somehow makes working endlessly acceptable.\nWhy this keeps happening Social media compresses complex life decisions into simple financial equations.\nMore effort → more money → better life.\nIt sounds neat.\nBut the equation ignores three realities that money cannot override.\nThe first is biology.\nNo one can work 24/7. Humans need sleep, recovery, mental space, and health. Ignore those long enough and performance collapses anyway.\nThe second is human need.\nPeople require relationships, exploration, curiosity, rest, and joy. These are not luxuries; they are structural components of a meaningful life.\nThe third is time.\nLife experience is not a renewable resource.\nYears spent exclusively pursuing money cannot be purchased back later.\nThe tempting but wrong response The easy reaction is to treat the trade purely economically.\n“If the pay is high enough, the sacrifice is worth it.”\nBut that logic quietly converts life into a narrow transaction.\nTime becomes a commodity.\nExperience becomes optional.\nIdentity collapses into productivity.\nAt that point, money isn’t serving life anymore.\nLife is serving money.\nA better way to think about it Ambition is not the problem.\nPeriods of intense effort are often necessary to build anything meaningful.\nBut intensity must be temporary and purposeful.\nThere is a difference between working very hard and structuring your life around nothing but work.\nThe first builds capability.\nThe second gradually erases everything else that makes success worth having.\nWhat changes when you apply this You stop measuring decisions only by income.\nYou start asking questions that include the full cost:\nWhat experiences will disappear if I follow this path? What relationships will weaken? What parts of life will I never revisit? Money becomes a tool for freedom.\nNot a justification for surrendering it.\nConcrete example There’s an old story about a fisherman and a businessman.\nA businessman visits a small coastal village and notices a fisherman returning with a few beautiful fish. Curious, he asks how long it took.\n“Not long,” the fisherman replies.\nThe businessman is puzzled. “Why don’t you stay out longer and catch more?”\nThe fisherman explains that he catches enough to support his family. The rest of the day he spends with his children, talks with friends in the village, and plays guitar in the evening.\nThe businessman smiles and begins outlining a grand plan.\n“You should fish longer,” he says. “Then you could buy a bigger boat. Eventually several boats. Then a fleet. You’d build a company, expand internationally, and sell the business for millions.”\nThe fisherman asks, “And then what?”\nThe businessman replies confidently:\n“Then you could retire. Move to a quiet village. Fish a little in the morning. Spend time with your family. Talk with friends. Play music in the evening.”\nThe fisherman looks at him calmly.\n“That’s what I’m doing now.”\nThe story sounds simple, even naive. But it exposes a powerful illusion.\nMany people are chasing a future lifestyle that resembles the life they already have — except they sacrifice decades to get there.\nWhat I’ve learned the hard way Ambition can quietly narrow your world.\nWhen you’re inside the grind, it feels justified.\nPerspective tends to arrive later, only when the years have already been spent.\nA rule of thumb If the only justification for a life decision is money, you probably haven\u0026rsquo;t calculated the real cost.\n","permalink":"https://borggrech.com/thoughts/illusion-money-can-buy-life/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: people claim they would happily work 24/7 if the salary were high enough\u003c/li\u003e\n\u003cli\u003eWhy this keeps appearing: social media reduces life decisions to financial trade-offs\u003c/li\u003e\n\u003cli\u003eThe flawed response: believing unlimited work is acceptable if the compensation is high enough\u003c/li\u003e\n\u003cli\u003eThe better framing: money cannot override biological limits, human needs, or lost life experience\u003c/li\u003e\n\u003cli\u003eWhat changes when you treat time and life experience as the real scarce resources\u003c/li\u003e\n\u003cli\u003eExample: the difference between temporary intensity and permanent sacrifice\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eYou see this comment everywhere.\u003c/p\u003e","title":"The Illusion That Money Can Buy Your Entire Life"},{"content":"Outline The tension: some wealthy people don’t appear analytically exceptional Why this perception persists: we overvalue visible intellect The flawed response: dismissing success as luck or connections The better framing: social intelligence is leverage What changes when you treat relational skill as strategic capital Example: the technically average operator who outperforms the brilliant loner Open with the real problem You’ve probably thought it.\nYou meet someone wealthy. Influential. Financially free.\nAnd quietly, you think: “They’re not that sharp.”\nThey’re not the best analyst in the room.\nThey’re not the deepest thinker.\nThey’re not always technically impressive.\nYet they compound.\nWhy this keeps happening We’ve been trained to recognize visible intelligence.\nAnalytical speed.\nTechnical mastery.\nStructured reasoning.\nThese are measurable. They’re rewarded in school. They’re easy to admire.\nBut markets reward something else too: access, persuasion, trust, timing.\nRelational fluency creates opportunity flow.\nIt attracts capital. It attracts forgiveness. It attracts information earlier.\nThat compounds quietly.\nThe tempting but wrong response The easy reaction is dismissal.\n“They just know people.”\n“They got lucky.”\n“They were in the right room.”\nBut being in the right room consistently is not an accident.\nMaintaining trust across rooms is work.\nInfluence is infrastructure.\nIgnoring that doesn’t reduce its power.\nA better way to think about it Intelligence isn’t only about solving problems.\nIt’s about moving systems.\nAnd systems include people.\nSome people optimize for depth.\nOthers optimize for alignment, trust, and optionality.\nBoth are forms of intellect.\nOne compounds through logic.\nThe other compounds through leverage.\nWhat changes when you apply this You stop underestimating soft power.\nYou start investing in:\nclarity in communication follow-through reputation consistency usefulness before visibility You realize social capital is not fluff. It’s durability.\nConcrete example Two founders start at the same time.\nFounder A builds a technically superior product but avoids networking, rarely follows up, and under-communicates wins.\nFounder B builds something solid, not exceptional — but calls people back, makes introductions, shares credit, and maintains trust.\nFive years later, Founder B has better capital access, stronger partnerships, and easier distribution.\nNot because of luck.\nBecause trust compounds faster than isolation.\nWhat I’ve learned the hard way Early in my career, I believed competence alone would carry everything.\nIt carries far but not everywhere.\nRelationships determine how far competence travels.\nA rule of thumb If people consistently want to work with you again, you are compounding something more powerful than intellect alone.\n","permalink":"https://borggrech.com/thoughts/social-skill-is-intelligence/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: some wealthy people don’t appear analytically exceptional\u003c/li\u003e\n\u003cli\u003eWhy this perception persists: we overvalue visible intellect\u003c/li\u003e\n\u003cli\u003eThe flawed response: dismissing success as luck or connections\u003c/li\u003e\n\u003cli\u003eThe better framing: social intelligence is leverage\u003c/li\u003e\n\u003cli\u003eWhat changes when you treat relational skill as strategic capital\u003c/li\u003e\n\u003cli\u003eExample: the technically average operator who outperforms the brilliant loner\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eYou’ve probably thought it.\u003c/p\u003e\n\u003cp\u003eYou meet someone wealthy. Influential. Financially free.\u003c/p\u003e","title":"Social Skill Is Intelligence (Even When It Doesn’t Look Like It)"},{"content":"Outline The tension: ideas are abundant, execution is scarce Why inaction persists: fear, diffusion of ownership, perfectionism The flawed response: waiting for clarity before acting The better framing: bias toward reversible motion What changes when movement becomes systemic Example: the prototype that never shipped vs the imperfect one that did Open with the real problem Most people don’t lack ideas.\nThey lack motion.\nSome of the most interesting ideas I’ve heard were shared quietly: never documented, never tested, never released.\nThey didn’t fail.\nThey expired.\nWhy this keeps happening Inaction feels safe.\nIf you don’t publish, you can’t be criticized.\nIf you don’t ship, you can’t be wrong.\nIf you don’t test, you can preserve the illusion of potential.\nOrganizations amplify this.\nUnclear ownership means no one moves.\nConsensus culture delays momentum.\nPerfection becomes protection.\nThe tempting but wrong response The common justification is prudence.\n“We need more clarity.”\n“We need full alignment.”\n“Timing isn’t right.”\nSometimes that’s true.\nOften it’s hesitation disguised as professionalism.\nClarity is often earned through contact with reality.\nA better way to think about it Bias toward reversible action.\nNot recklessness.\nReversible movement.\nSmall pilots.\nLimited exposure.\nTight feedback loops.\nMomentum reduces fear because it replaces imagination with data.\nWhat changes when you apply this Decisions accelerate.\nEnergy shifts from debate to iteration.\nConfidence grows from exposure, not speculation.\nAnd teams stop romanticizing ideas they’ve never tested.\nConcrete example Two teams explore a new AI-enabled workflow.\nTeam A debates architecture for three months.\nTeam B builds a basic internal version in two weeks and tests with five users.\nSix months later, Team B has a refined system.\nTeam A has a polished strategy deck.\nNeither team lacked intelligence.\nOne lacked motion.\nWhat I’ve learned the hard way The regret is rarely failure.\nIt’s hesitation.\nA simple checklist Is this action reversible? What is the smallest viable test? Who owns the decision to move? What are we actually afraid of? What data would reduce this fear? ","permalink":"https://borggrech.com/thoughts/graveyard-of-unshipped-ideas/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: ideas are abundant, execution is scarce\u003c/li\u003e\n\u003cli\u003eWhy inaction persists: fear, diffusion of ownership, perfectionism\u003c/li\u003e\n\u003cli\u003eThe flawed response: waiting for clarity before acting\u003c/li\u003e\n\u003cli\u003eThe better framing: bias toward reversible motion\u003c/li\u003e\n\u003cli\u003eWhat changes when movement becomes systemic\u003c/li\u003e\n\u003cli\u003eExample: the prototype that never shipped vs the imperfect one that did\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eMost people don’t lack ideas.\u003c/p\u003e\n\u003cp\u003eThey lack motion.\u003c/p\u003e\n\u003cp\u003eSome of the most interesting ideas I’ve heard were shared quietly: never documented, never tested, never released.\u003c/p\u003e","title":"The Graveyard of Unshipped Ideas"},{"content":"Outline The tension: answers are instant, depth is rare Why this happens: speed incentives and output culture The flawed response: outsourcing thinking to tools The better framing: protect pondering as design time What changes when you slow down strategically Example: strategy drafted with and without internal reflection Open with the real problem We can get answers instantly now.\nAny question. Any format.\nThat’s extraordinary.\nIt also shortens the time we spend with the question.\nWhy this keeps happening Output is visible.\nReflection is invisible.\nAI reduces friction so dramatically that we skip the uncomfortable pause.\nWe move from question to answer without letting the tension work on us.\nDepth requires friction.\nSpeed removes friction.\nThe tempting but wrong response The easy path is default generation.\nAsk. Copy. Paste. Move on.\nThe output sounds plausible. So we trust it.\nBut thinking is not only producing words.\nIt is wrestling with uncertainty.\nA better way to think about it Treat pondering as protected design time.\nNot nostalgia.\nDesign.\nBefore asking the machine, outline your assumptions.\nBefore generating, clarify constraints.\nWhen you return to the tool, your prompts are sharper because your thinking is sharper.\nWhat changes when you apply this Arguments tighten.\nDecisions calm down.\nReactivity decreases.\nYou regain authorship over your thinking.\nConcrete example Drafting a strategic memo:\nOption A: prompt AI immediately for a full strategy.\nOption B: spend 30 minutes outlining risks, constraints, tradeoffs: then use AI to stress-test.\nOption B feels slower.\nIt produces better judgment.\nWhat I’ve learned the hard way Speed can simulate progress.\nDepth compounds more quietly.\nA rule of thumb If the decision matters, sit with the question before you automate the answer.\n","permalink":"https://borggrech.com/thoughts/lost-art-of-pondering/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: answers are instant, depth is rare\u003c/li\u003e\n\u003cli\u003eWhy this happens: speed incentives and output culture\u003c/li\u003e\n\u003cli\u003eThe flawed response: outsourcing thinking to tools\u003c/li\u003e\n\u003cli\u003eThe better framing: protect pondering as design time\u003c/li\u003e\n\u003cli\u003eWhat changes when you slow down strategically\u003c/li\u003e\n\u003cli\u003eExample: strategy drafted with and without internal reflection\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eWe can get answers instantly now.\u003c/p\u003e\n\u003cp\u003eAny question. Any format.\u003c/p\u003e\n\u003cp\u003eThat’s extraordinary.\u003c/p\u003e\n\u003cp\u003eIt also shortens the time we spend with the question.\u003c/p\u003e","title":"The Lost Art of Pondering"},{"content":"Outline The tension: Energym feels absurd, yet unsettlingly plausible Why this keeps resurfacing: AI abundance shifts scarcity from production to meaning The flawed response: dismissing the satire or fearing it literally The better framing: energy and attention are the real constraints What changes when we treat engagement as a design problem Example: redesigning AI systems around dignity and purpose, not just efficiency Open with the real problem The viral “Energym” video shows an imagined future where humans power AI systems through physical workouts.\nIt’s satire.\nBut it spreads because it feels emotionally plausible.\nNot because we expect gym bikes to run datacenters.\nBecause we intuit something deeper:\nIf AI creates abundance, what do humans do with their energy?\nAnd what powers the system that creates that abundance?\nWhy this keeps happening We talk about AI as if it only increases output.\nBut abundance changes scarcity.\nWhen production becomes cheap, attention becomes scarce.\nWhen answers are instant, meaning becomes harder to earn.\nA world of abundant digital capability does not automatically create engaged humans.\nIt can create passive ones.\nThe Energym satire works because it compresses two fears into one image:\nAI’s growing physical energy demands. Humans needing structured engagement to remain meaningfully involved. In The Matrix, humans become batteries.\nIn Energym, humans become participants in the system.\nThe question underneath both is the same:\nIf machines do the work, what keeps humans engaged in life?\nThe tempting but wrong response The first lazy reaction is dismissal.\n“It’s just a meme.”\nThe second is dystopian panic.\n“We’re heading toward exploitation.”\nBoth avoid the harder question:\nHow do we design a world of abundance that still requires human agency?\nIf AI reduces friction everywhere, humans don’t automatically flourish.\nThey drift.\nEngagement doesn’t appear by accident.\nIt must be designed.\nA better way to think about it Energym, intentionally or not, touches a profound idea:\nIn a world with abundant resources, humans still need a structured way to direct attention.\nAttention is the new scarcity.\nEnergy powers machines.\nAttention powers meaning.\nIf production becomes abundant, the challenge shifts from “How do we survive?” to “How do we remain engaged?”\nThat is not a technical problem. It’s a design problem.\nOrganizations face it already.\nWhen AI increases productivity, the real risk isn’t unemployment alone.\nIt’s disengagement.\nPeople need:\nclear ownership visible contribution feedback loops meaningful constraint Remove all constraint and you don’t create freedom.\nYou create drift.\nWhat changes when you apply this You stop designing AI systems only for efficiency.\nYou start designing them for dignity.\nThat means:\nhumans remain accountable owners of outcomes AI drafts, humans decide outputs are reviewed, not blindly trusted roles evolve instead of evaporate You treat attention as precious.\nYou protect it.\nBecause a disengaged human is not liberated.\nThey’re sidelined.\nConcrete example Consider a company that automates 70% of its internal reporting with AI.\nOption A: Remove analysts. Let dashboards update automatically.\nProductivity rises. Engagement drops. Insight declines because no one feels responsible.\nOption B: AI prepares drafts and highlights anomalies. Analysts interpret, challenge, and contextualize.\nOutput speed increases. Ownership remains human. Attention remains active.\nThe difference isn’t technology.\nIt’s design.\nWhat I’ve learned the hard way Efficiency without agency erodes morale quietly.\nPeople don’t only need income.\nThey need involvement.\nA rule of thumb In a world of abundance, design for engagement.\nBecause attention, not energy: is the ultimate fuel of human systems.\n","permalink":"https://borggrech.com/thoughts/energym-isnt-about-the-gym/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe tension: Energym feels absurd, yet unsettlingly plausible\u003c/li\u003e\n\u003cli\u003eWhy this keeps resurfacing: AI abundance shifts scarcity from production to meaning\u003c/li\u003e\n\u003cli\u003eThe flawed response: dismissing the satire or fearing it literally\u003c/li\u003e\n\u003cli\u003eThe better framing: energy and attention are the real constraints\u003c/li\u003e\n\u003cli\u003eWhat changes when we treat engagement as a design problem\u003c/li\u003e\n\u003cli\u003eExample: redesigning AI systems around dignity and purpose, not just efficiency\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eThe viral “Energym” video shows an imagined future where humans power AI systems through physical workouts.\u003c/p\u003e","title":"Energym Isn’t About the Gym. It’s About Power, Attention, and Human Purpose."},{"content":"What this playbook is for For teams adopting AI quickly and feeling two fears at once:\n“We’re moving too slow if we don’t use it.” “We’re taking risks we can’t see.” This playbook is designed for small teams who need speed and reliability.\nOutcomes you should expect Clear boundaries on where AI is allowed. Human accountability remains visible. Fewer quality incidents caused by unchecked output. Faster adoption because people feel safe using it. The operating principle AI does not remove accountability. It changes what the human is accountable for.\nGuardrails are not heavy policy. Guardrails are design constraints:\nwhere AI can be used what must be verified who owns the outcome how incidents are handled The artifacts (copy/paste) AI Use Boundaries (one page)\nAllowed workflows Not allowed (sensitive data, final approvals, legal commitments) Required verification steps Human-in-the-loop Checklist\nWhat must be checked (facts, numbers, tone, commitments) What must be cited (sources, assumptions) AI Incident Note (one page)\nWhat happened | Impact | Root cause | Guardrail update | Owner Prompt + Output Log (optional)\nNot for surveillance; for learning and repeatability Step-by-step implementation Pick 3 workflows where AI helps but risk is manageable. Define verification: what must the human check before output ships? Assign ownership: one person owns the final outcome. Define “no-go zones” (data and decisions that must remain human). Create an incident path: when output causes harm, document and update guardrails. Review monthly: expand allowed workflows only when reliability is proven. Cadence (how you keep it alive) Weekly:\nQuick review: where did AI help, where did it create risk? Monthly:\nUpdate allowed workflows. Review AI incidents and adjust verification. Refresh boundaries as the business changes. Metrics that prove it’s working Number of AI-assisted outputs shipped. Quality incidents attributable to AI output. Time saved in selected workflows. Adoption rate (how many people use it confidently). Rework rate (output revised after shipping). Common failure modes (and fixes) Over-restriction kills adoption. Fix: allow low-risk workflows first. No verification leads to incidents. Fix: simple checklist tied to workflow. No owner means blame storms. Fix: one DRI per output type. Silent failures repeat. Fix: one-page incident notes + guardrail updates. Example: what this looks like in a real week A team allowed AI for:\ndrafting customer emails (human reviews before send) summarizing meetings into decisions (decision log required) first-pass SOP drafts (owner approves) They banned AI for:\nlegal commitments pricing promises sensitive personal data Incidents dropped because verification was explicit, not assumed.\nQuick start checklist Choose 3 allowed workflows. Write a one-page boundary doc. Add a verification checklist. Assign a DRI for each output type. Start logging incidents when they happen, without blame. ","permalink":"https://borggrech.com/thoughts/ai-safety-net-playbook/","summary":"\u003ch2 id=\"what-this-playbook-is-for\"\u003eWhat this playbook is for\u003c/h2\u003e\n\u003cp\u003eFor teams adopting AI quickly and feeling two fears at once:\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e“We’re moving too slow if we don’t use it.”\u003c/li\u003e\n\u003cli\u003e“We’re taking risks we can’t see.”\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThis playbook is designed for small teams who need speed \u003cem\u003eand\u003c/em\u003e reliability.\u003c/p\u003e\n\u003ch2 id=\"outcomes-you-should-expect\"\u003eOutcomes you should expect\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eClear boundaries on where AI is allowed.\u003c/li\u003e\n\u003cli\u003eHuman accountability remains visible.\u003c/li\u003e\n\u003cli\u003eFewer quality incidents caused by unchecked output.\u003c/li\u003e\n\u003cli\u003eFaster adoption because people feel safe using it.\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"the-operating-principle\"\u003eThe operating principle\u003c/h2\u003e\n\u003cp\u003eAI does not remove accountability.\nIt changes what the human is accountable for.\u003c/p\u003e","title":"The AI Safety Net Playbook: Guardrails Without Bureaucracy"},{"content":"What this playbook is for This is for founders and operators who feel the same pattern repeating: Everything is “fast” until it’s suddenly messy.\nSymptoms:\ndecisions happen in a meeting, then disappear people redo work because the latest version is unclear meetings multiply because nothing is written down ownership is implied, not explicit Outcomes you should expect Fewer meetings because decisions and plans exist as artifacts. Less rework because “done” is defined. Faster onboarding because new people can read the system. Higher trust because ownership and expectations are explicit. The operating principle You don’t scale with more process. You scale with a few thin artifacts that make work legible.\nThin means:\nsmall enough to maintain clear enough to reduce debate structured enough to survive growth The goal is not documentation. The goal is coordination with fewer conversations.\nThe artifacts (copy/paste) Copy/paste these as simple docs (one page each):\nOwner Map (one screen) Area → Directly Responsible Individual (DRI) → backup → escalation Decision Log (the antidote to meeting amnesia) Decision | Date | DRI | Options considered | Why | Next review date Weekly Plan (per team) Top 3 outcomes | Risks | Dependencies | What we will not do Work Brief (for non-trivial work) Problem | Constraints | Definition of done | Risks | Owner | Due date Postmortem Note (when things break) What happened | Impact | Root cause | Fix | Prevention | Owner Step-by-step implementation Day 1: Install the artifacts\nCreate the Owner Map (even if imperfect). Create a Decision Log and commit to using it. Agree that non-trivial work needs a Work Brief. Week 1: Change the rules of communication\nIf it’s a decision, it goes in the Decision Log. If it’s work with risk, it needs a Work Brief. If someone asks “what are we doing?”, point to the Weekly Plan. Week 2: Remove meetings that exist only to compensate\nCancel any meeting whose output is “alignment.” Replace it with a Weekly Plan + Decision Log review. Week 3: Make ownership sharp\nEvery outcome has one DRI. Committees advise. DRIs decide. Week 4: Lock in the habit\nRun a short retro: which artifact is ignored? Fix the artifact, not the people. Cadence (how you keep it alive) Weekly (30-45 min per team):\nUpdate Weekly Plan (top 3 outcomes + risks). Review Decision Log: what needs a revisit? Monthly (30 min):\nRefresh Owner Map. Review the top 3 recurring failure patterns and add a guardrail. Metrics that prove it’s working Meeting hours per person per week (should fall). Rework incidents (same work done twice). Cycle time for decisions (question → recorded decision). Onboarding time to first independent outcome. Number of “where is this?” messages (a proxy for missing artifacts). Common failure modes (and fixes) Artifacts become “busywork.” Fix: reduce fields until it’s one screen. No one uses the Decision Log. Fix: leader models it; review it weekly. Ownership stays vague. Fix: force one DRI; accept discomfort. People keep meeting out of habit. Fix: require a written agenda + expected artifact output. Everything becomes a Work Brief. Fix: only for non-trivial work (risk, dependency, customer impact). Example: what this looks like in a real week Monday:\nTeam updates Weekly Plan in 20 minutes. Two decisions are logged. Tuesday:\nA new initiative starts with a Work Brief. Stakeholders comment asynchronously; DRI decides. Thursday:\nA production issue triggers a one-page postmortem. The prevention step becomes a new guardrail in the Work Brief template. Friday:\nNo “status meeting.” The Weekly Plan is the status. Quick start checklist Create the Owner Map today. Start the Decision Log today. Replace one recurring meeting with a Weekly Plan review. Require Work Briefs for work that can hurt customers. Measure meeting hours next week and cut 10%. ","permalink":"https://borggrech.com/thoughts/thin-process-playbook/","summary":"\u003ch2 id=\"what-this-playbook-is-for\"\u003eWhat this playbook is for\u003c/h2\u003e\n\u003cp\u003eThis is for founders and operators who feel the same pattern repeating:\nEverything is “fast” until it’s suddenly messy.\u003c/p\u003e\n\u003cp\u003eSymptoms:\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003edecisions happen in a meeting, then disappear\u003c/li\u003e\n\u003cli\u003epeople redo work because the latest version is unclear\u003c/li\u003e\n\u003cli\u003emeetings multiply because nothing is written down\u003c/li\u003e\n\u003cli\u003eownership is implied, not explicit\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"outcomes-you-should-expect\"\u003eOutcomes you should expect\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eFewer meetings because decisions and plans exist as artifacts.\u003c/li\u003e\n\u003cli\u003eLess rework because “done” is defined.\u003c/li\u003e\n\u003cli\u003eFaster onboarding because new people can read the system.\u003c/li\u003e\n\u003cli\u003eHigher trust because ownership and expectations are explicit.\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"the-operating-principle\"\u003eThe operating principle\u003c/h2\u003e\n\u003cp\u003eYou don’t scale with more process.\nYou scale with a few \u003cem\u003ethin\u003c/em\u003e artifacts that make work legible.\u003c/p\u003e","title":"The Thin Process Playbook: Just Enough Structure to Scale"},{"content":"Outline The core problem or tension you’re addressing (grounded in reality) Why this problem keeps showing up (systemic, not personal) The common but flawed response most teams default to The better framing or operating principle What changes when this principle is applied Example: a concrete, real-world scenario Open with the real problem Leaders say they want autonomy. Teams say they want trust.\nBut under pressure, control returns. Suddenly there are more check-ins, more approvals, more monitoring.\nEveryone feels insulted. No one feels safe.\nWhy this keeps happening Trust breaks when outcomes become unpredictable. And unpredictability usually comes from systems:\nunclear commitments changing priorities hidden dependencies no owner for quality People don’t lose trust because someone is imperfect. They lose trust because they can’t forecast reality.\nThe tempting but wrong response The flawed response is surveillance. More tracking, more reporting, more status.\nThat often reduces trust further. People optimize for appearances.\nThe other flawed response is blind trust. That ignores real risk and repeats mistakes.\nA better way to think about it Treat trust as a ledger of commitments.\nA commitment is not “I’ll try.” It is:\na specific outcome a date a proof artifact an owner Trust grows when commitments are consistently kept. Autonomy becomes earned, not granted by mood.\nWhat changes when you apply this When trust becomes a ledger:\nautonomy becomes easier to grant micromanagement decreases expectations become explicit underperformance becomes diagnosable The team stops debating trust and starts building it.\nConcrete example A manager felt a team member was “not responsive.” They replaced the debate with commitments:\nweekly outcomes posted Monday proof artifacts linked Friday After two weeks, the signal was clear: Either outcomes were delivered (trust increased) or they weren’t (system or capability issue).\nA simple checklist / rule of thumb A simple checklist Are commitments explicit or implied? Is there a proof artifact for each outcome? Are dependencies visible early? Are priorities stable enough to keep commitments? Are misses treated as system signals before moral failures? Rule of thumb Autonomy follows reliability. Reliability follows clarity.\n","permalink":"https://borggrech.com/thoughts/trust-ledger/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe core problem or tension you’re addressing (grounded in reality)\u003c/li\u003e\n\u003cli\u003eWhy this problem keeps showing up (systemic, not personal)\u003c/li\u003e\n\u003cli\u003eThe common but flawed response most teams default to\u003c/li\u003e\n\u003cli\u003eThe better framing or operating principle\u003c/li\u003e\n\u003cli\u003eWhat changes when this principle is applied\u003c/li\u003e\n\u003cli\u003eExample: a concrete, real-world scenario\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eLeaders say they want autonomy.\nTeams say they want trust.\u003c/p\u003e\n\u003cp\u003eBut under pressure, control returns.\nSuddenly there are more check-ins, more approvals, more monitoring.\u003c/p\u003e","title":"The Trust Ledger: How Autonomy Is Earned (and Lost)"},{"content":"Outline Drama is often a symptom of unclear ownership and late feedback Calm teams write more and interrupt less A high bar is maintained through systems: criteria, review, and learning loops Treat incidents as system signals, not personal failures Leaders set tone through what they reward and what they tolerate Example: reducing meeting load while improving quality Drama is expensive, even when it looks productive Some teams confuse intensity with effectiveness.\nFast talking, urgent meetings, late-night heroics. It can feel like commitment. It can also be a sign that the system is failing: unclear priorities, unclear ownership, and late discovery of problems.\nCalm execution doesn’t mean “slow.” It means the team is not surprised every day. Work moves with fewer spikes. Feedback comes early. Decisions are explicit.\nAI can reduce some busywork, but it won’t fix a culture of urgency-as-identity. That’s a leadership choice.\nIf you want to attract high-quality operators, calm execution is a competitive advantage.\nCalm teams write more and interrupt less One practical marker: calm teams use writing as the default.\nThey write one-pagers for decisions. They write short runbooks for repeated work. They write incident notes that become improvements.\nInterruptions are expensive. They fragment attention and increase errors. Calm teams protect deep work and reserve interruptions for true emergencies.\nAI helps here by accelerating drafts and summaries. But the discipline is human: deciding that written artifacts are the primary interface for work.\nIf your team’s knowledge lives in people’s heads and chat threads, urgency will dominate. If it lives in lightweight documents, calm becomes possible.\nA high bar is a system, not a mood A “high bar” is not a vibe. It’s a set of visible standards and feedback loops.\nThe system includes:\nacceptance criteria for outputs examples of excellent and unacceptable work review mechanisms (sampling is often enough) clear ownership for outcomes learning loops that update the standard Without systems, standards become personality-driven. People guess what the leader wants. That creates anxiety and politics.\nWith systems, people can self-correct. They can ship confidently. They can disagree without fear because the criteria are shared.\nAI raises the need for this because output is easier. The bar must be anchored in explicit criteria, not in “I’ll know it when I see it.”\nTreat incidents as signals, not moral failures Calm teams handle failure differently.\nThey don’t pretend failure won’t happen. They assume it will. They treat incidents as signals about the system: where ownership was unclear, where review was weak, where assumptions were wrong.\nThis doesn’t mean no accountability. It means accountability is applied at the right level: fix the system so the incident is less likely to recur.\nThe fastest way to destroy calm is to turn every incident into blame. People will hide problems until they become large. That creates drama.\nThe fastest way to build calm is to normalize early detection and transparent reporting.\nLeaders set tone by what they reward and tolerate Culture is not what you say. It’s what you reward.\nIf you praise heroics, you’ll get heroics. If you reward clear planning, thoughtful writing, and steady delivery, you’ll get calm.\nAlso: what you tolerate becomes the standard.\nIf you tolerate unclear ownership, you’ll get blame loops. If you tolerate sloppy work, you’ll get drift. If you tolerate disrespect, you’ll lose strong people.\nCalm execution is not passive. It is disciplined. It is demanding in a quiet way.\nConcrete example: fewer meetings, better quality Imagine a team with a recurring pattern: many meetings, frequent “urgent” issues, and inconsistent delivery.\nYou introduce three changes:\nDecision one-pagers replace most meetings. Each has a decider and a deadline. A sampling review loop for customer-facing outputs and production changes. A short incident note format: what happened, why, what changed, owner, next review. Within a month, meeting time drops. People spend more time on real work. Quality improves because drift is caught earlier. Urgency decreases because fewer problems are discovered late.\nThe team becomes calmer, not because they care less, but because the system supports them.\nA simple checklist Where is drama compensating for unclear ownership or late feedback? Are decisions written with a decider, deadline, and recorded outcome? Do we have explicit standards and examples for “good”? Do we run sampling review to detect drift early? Do we treat incidents as system signals with clear follow-up owners? What behaviors do leaders reward: heroics or reliability? ","permalink":"https://borggrech.com/thoughts/calm-execution-high-bar/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eDrama is often a symptom of unclear ownership and late feedback\u003c/li\u003e\n\u003cli\u003eCalm teams write more and interrupt less\u003c/li\u003e\n\u003cli\u003eA high bar is maintained through systems: criteria, review, and learning loops\u003c/li\u003e\n\u003cli\u003eTreat incidents as system signals, not personal failures\u003c/li\u003e\n\u003cli\u003eLeaders set tone through what they reward and what they tolerate\u003c/li\u003e\n\u003cli\u003eExample: reducing meeting load while improving quality\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"drama-is-expensive-even-when-it-looks-productive\"\u003eDrama is expensive, even when it looks productive\u003c/h2\u003e\n\u003cp\u003eSome teams confuse intensity with effectiveness.\u003c/p\u003e","title":"Calm Execution: Building a High Bar Without the Drama"},{"content":"Outline The core problem or tension you’re addressing (grounded in reality) Why this problem keeps showing up (systemic, not personal) The common but flawed response most teams default to The better framing or operating principle What changes when this principle is applied Example: a concrete, real-world scenario Open with the real problem You feel it in the room. A good idea appears, and before it takes its first breath, the process kills it.\nSomeone says: “We can’t do that because we don’t do it that way.” No one is trying to be difficult. They’re protecting the system that used to keep things safe.\nWhy this keeps happening Processes are often created after pain. A failure happens, so a rule is born.\nThe problem is that the world changes faster than the process. So teams keep enforcing constraints that no longer match reality.\nThis keeps happening when:\nincentives reward compliance over outcomes people are punished for risk but not for stagnation decision rights are unclear, so process becomes a shield The tempting but wrong response The tempting response is to declare war on process. “Move fast. No red tape.”\nThat often creates chaos. Chaos forces control. And control eventually becomes process again.\nThe other tempting response is to accept the process and work around it quietly. That creates cynicism and slow decay.\nA better way to think about it Treat process as a design choice that must earn its keep.\nA better framing:\nKeep the process that protects customers and quality. Remove the process that protects egos and habits. Create small “edge loops” where experimentation is allowed. Creativity needs constraints. It just needs the right ones: clarity, ownership, and fast feedback.\nWhat changes when you apply this When you de-calcify process:\nteams propose ideas because they believe action is possible learning speed increases (small experiments, short cycles) ownership becomes real because decisions are explicit risk becomes manageable because experiments are bounded Small companies win by learning faster than big companies can approve.\nConcrete example A team wanted to try a new onboarding flow. The old process required 6 approvals and a quarterly roadmap slot.\nThey introduced an edge loop:\nexperiment budget: 2 weeks customer impact boundary: only new signups success metric: activation rate DRI empowered to decide They shipped a small version, measured results, and either scaled or killed it. No heroics. No theatre. Just learning.\nA simple checklist / rule of thumb A simple checklist What pain did this process originally solve? Does that pain still exist at the same level? Who benefits from keeping the process (customers, team, or bureaucracy)? Can we create a bounded experiment instead of a full change request? Who is the DRI who can say yes? Rule of thumb If a process can’t explain the customer it protects, it’s probably protecting the past.\n","permalink":"https://borggrech.com/thoughts/yesterdays-rules-suffocate-todays-creativity/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe core problem or tension you’re addressing (grounded in reality)\u003c/li\u003e\n\u003cli\u003eWhy this problem keeps showing up (systemic, not personal)\u003c/li\u003e\n\u003cli\u003eThe common but flawed response most teams default to\u003c/li\u003e\n\u003cli\u003eThe better framing or operating principle\u003c/li\u003e\n\u003cli\u003eWhat changes when this principle is applied\u003c/li\u003e\n\u003cli\u003eExample: a concrete, real-world scenario\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eYou feel it in the room.\nA good idea appears, and before it takes its first breath, the process kills it.\u003c/p\u003e","title":"Process Calcification: When Yesterday’s Rules Suffocate Today’s Creativity"},{"content":"What this playbook is for This is for teams working across countries and time zones who want remote work to feel:\ncalm (not chaotic) fair (not proximity-biased) high-output (not performative) Especially relevant as EU collaboration increases and cross-border teams become normal.\nOutcomes you should expect Clear expectations: people know what “good” looks like. Fewer meetings, more artifacts. Predictable overlap windows for customers and colleagues. Trust based on evidence (outputs), not surveillance. Reduced anxiety about “are they working?” The operating principle Remote work fails when teams try to copy the office. It succeeds when teams shift from time-based control to outcome-based clarity.\nTrust is not blind. Trust is a system:\nvisible commitments observable outputs clear decision rights fast feedback when something slips The artifacts (copy/paste) Copy/paste these:\nWorking Agreement (one page) Core overlap hours (e.g., 10:00-14:00 CET) Response-time expectations (e.g., 4 hours within overlap) Meeting rules (written-first, invite-only) Outcome Board (weekly) Each person: Top 2 outcomes, due dates, proof artifact link Decision Log Decisions recorded with DRI + rationale Customer Coverage Map Who covers which hours, escalation path Trust Ledger (lightweight) Commitments made → commitments kept (signals reliability) Not a surveillance tool; a reliability tool Step-by-step implementation Week 1: Install clarity\nSet overlap hours (start with 3-4 hours/day). Create Outcome Board and use it immediately. Agree on response times inside and outside overlap. Week 2: Reduce meetings\nRequire written agendas and expected artifact outputs. Replace “status” with written updates linked to outcomes. Week 3: Make ownership explicit\nOne DRI per outcome. Define escalation: when do we interrupt overlap hours? Week 4: Build trust by proof\nReview outcomes weekly. Celebrate reliability. Address misses as system issues first (unclear scope, too many dependencies). Cadence (how you keep it alive) Daily (async):\nUpdate Outcome Board when something changes. Weekly (45 min):\nReview outcomes: shipped, blocked, slipped. Review Decision Log. Agree next week’s top outcomes. Monthly (30 min):\nRevisit overlap hours and customer coverage. Fix one recurring coordination failure. Metrics that prove it’s working Outcome completion rate (weekly). Average time-to-response during overlap. Meeting hours per person. Customer escalations outside coverage hours. “Work visibility” messages (should drop as artifacts improve). Common failure modes (and fixes) Remote becomes always-on. Fix: define quiet hours and escalation rules. People game the system. Fix: measure outcomes, not presence. Too much async causes drift. Fix: keep a small overlap window. Meetings creep back. Fix: cancel recurring meetings that don’t produce artifacts. Trust breaks after one miss. Fix: diagnose constraints; adjust scope and decision rights. Example: what this looks like in a real week A distributed team sets overlap 10:00-14:00 CET.\nMonday:\nEveryone posts top 2 outcomes with proof artifacts. Wednesday:\nA risk appears; DRI logs a decision and escalates only if needed. Friday:\nWeekly review: outcomes shipped, slips explained, next week committed. No one needs to ask “are you online?” The system answers it.\nQuick start checklist Set 3-4 overlap hours. Create an Outcome Board and use it tomorrow. Replace one status meeting with a written update. Start a Decision Log. Review outcomes weekly, not activity daily. ","permalink":"https://borggrech.com/thoughts/the-borderless-remote-playbook/","summary":"\u003ch2 id=\"what-this-playbook-is-for\"\u003eWhat this playbook is for\u003c/h2\u003e\n\u003cp\u003eThis is for teams working across countries and time zones who want remote work to feel:\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ecalm (not chaotic)\u003c/li\u003e\n\u003cli\u003efair (not proximity-biased)\u003c/li\u003e\n\u003cli\u003ehigh-output (not performative)\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eEspecially relevant as EU collaboration increases and cross-border teams become normal.\u003c/p\u003e\n\u003ch2 id=\"outcomes-you-should-expect\"\u003eOutcomes you should expect\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eClear expectations: people know what “good” looks like.\u003c/li\u003e\n\u003cli\u003eFewer meetings, more artifacts.\u003c/li\u003e\n\u003cli\u003ePredictable overlap windows for customers and colleagues.\u003c/li\u003e\n\u003cli\u003eTrust based on evidence (outputs), not surveillance.\u003c/li\u003e\n\u003cli\u003eReduced anxiety about “are they working?”\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"the-operating-principle\"\u003eThe operating principle\u003c/h2\u003e\n\u003cp\u003eRemote work fails when teams try to copy the office.\nIt succeeds when teams shift from \u003cstrong\u003etime-based control\u003c/strong\u003e to \u003cstrong\u003eoutcome-based clarity\u003c/strong\u003e.\u003c/p\u003e","title":"The Borderless Remote Playbook: Outcomes, Time Zones, and Trust (EU-Ready)"},{"content":"Outline AI makes output cheap; judgment becomes the scarce asset Look for problem framing, not just execution speed Test verification habits: can they catch confident mistakes? Evaluate tradeoff thinking: risk, cost, time, and second-order effects Hire for learning loops and written clarity Example: an interview exercise that reveals judgment quickly When output is cheap, the differentiator is decision quality AI can draft code, emails, and analysis. That doesn’t mean roles disappear. It means the skills shift.\nThe most valuable people will be those who can:\nframe problems clearly choose the right level of rigor verify claims and sources make tradeoffs without hand-waving communicate decisions in writing In other words: judgment.\nIf you keep hiring as if output volume is the primary skill, you’ll get people who can generate a lot—without reliably moving outcomes forward.\nProblem framing is the first test of judgment Strong operators don’t start with solutions. They start with a crisp frame:\nWhat are we trying to achieve? What constraints matter? What does success look like? What’s the smallest useful step? In interviews, many candidates jump to execution. They want to show competence. They produce answers quickly.\nYou want to see if they can slow down enough to frame the problem. AI will handle speed. Humans need to handle direction.\nA good sign is when someone asks clarifying questions that reveal they’re thinking about outcomes, not just tasks.\nVerification habits matter more than confidence AI increases the risk of “confident wrong.” The same is true for humans.\nYou want people who treat claims as checkable. They don’t say “I think.” They say “Here’s how I’d validate.”\nIn practice, you can test this by giving a candidate an AI-generated answer with subtle errors and asking them to review it.\nDo they accept it because it sounds good? Or do they interrogate assumptions, ask for sources, and propose verification steps?\nThis is not about being skeptical for its own sake. It’s about building systems that don’t collapse under speed.\nTradeoff thinking is what separates operators from technicians In real work, you rarely get to optimize one variable.\nYou trade speed for quality, cost for reliability, simplicity for flexibility. People with strong judgment can articulate tradeoffs explicitly.\nThey can say: “If we choose A, we gain speed but risk drift. If we choose B, we slow down but get auditability. Given our constraints, I recommend…”\nThis is especially important when AI proposes multiple options. AI can list tradeoffs. But it can’t own the choice. You need people who can.\nIn interviews, ask for past examples: “Tell me about a time you chose the ‘worse’ option and why.” The answer reveals maturity.\nLearning loops and written clarity are force multipliers In the AI era, tools change quickly. The people who thrive are those who learn continuously without drama.\nLook for:\ncomfort with feedback ability to update beliefs habit of writing things down ability to teach others and share patterns Written clarity is not just communication. It’s thinking. People who write well tend to think more clearly about constraints and outcomes.\nThat’s why writing-first teams scale better. And it’s why your hiring should value writing.\nConcrete example: an interview exercise for judgment A simple exercise:\nGive the candidate a short scenario: “A customer reports an issue. We have logs, some vague error messages, and a partial reproduction.”\nProvide an AI-drafted response plan that includes a few subtle problems: an overpromise, a missing verification step, a risky assumption.\nAsk the candidate to:\ncritique the plan propose a revised plan identify what they would verify first state what they would tell the customer and what they would not This reveals judgment quickly. Great candidates will:\nask for missing context identify risk boundaries propose verification steps communicate with calm precision Average candidates will focus on sounding smart.\nA simple checklist Are we hiring for judgment, or for output volume? Do candidates frame problems before solving them? Can they verify claims and catch confident mistakes? Do they articulate tradeoffs with real constraints? Can they write clearly and reason in public? Do they show learning loops: feedback → adjustment → improved approach? ","permalink":"https://borggrech.com/thoughts/hiring-for-judgment-ai-era/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eAI makes output cheap; judgment becomes the scarce asset\u003c/li\u003e\n\u003cli\u003eLook for problem framing, not just execution speed\u003c/li\u003e\n\u003cli\u003eTest verification habits: can they catch confident mistakes?\u003c/li\u003e\n\u003cli\u003eEvaluate tradeoff thinking: risk, cost, time, and second-order effects\u003c/li\u003e\n\u003cli\u003eHire for learning loops and written clarity\u003c/li\u003e\n\u003cli\u003eExample: an interview exercise that reveals judgment quickly\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"when-output-is-cheap-the-differentiator-is-decision-quality\"\u003eWhen output is cheap, the differentiator is decision quality\u003c/h2\u003e\n\u003cp\u003eAI can draft code, emails, and analysis. That doesn’t mean roles disappear. It means the skills shift.\u003c/p\u003e","title":"Hiring for Judgment in the AI Era"},{"content":"Outline The core problem or tension you’re addressing (grounded in reality) Why this problem keeps showing up (systemic, not personal) The common but flawed response most teams default to The better framing or operating principle What changes when this principle is applied Example: a concrete, real-world scenario Open with the real problem You open your calendar and it feels like a hostile takeover. Recurring meetings have multiplied, and somehow you still don’t feel aligned.\nPeople leave calls with action items, then schedule another call to “make sure it happens.” The work slows down, but the talking speeds up.\nWhy this keeps happening Meetings rarely multiply because people love meetings. They multiply because the system is missing something:\ndecisions are not recorded ownership is unclear “done” is not defined there’s no single source of truth When work is not legible, the only way to coordinate is to talk. So the org becomes a chat room with a payroll.\nThe tempting but wrong response The common response is to add “better meetings.” Agendas, tighter facilitation, more discipline.\nThat helps a bit, but it misses the point. Meetings are an expensive workaround for missing artifacts.\nAnother flawed response is to declare “async only” without building the written system. That just creates confusion in a different format.\nA better way to think about it Treat meetings as a last-mile tool, not the main operating system.\nA simple principle: If a meeting does not produce an artifact that changes future behavior, it’s mostly theatre.\nReplace “alignment” with artifacts:\nDecision log (what we decided and why) Owner map (who owns what) Work brief (what we’re doing and what done means) Weekly plan (top outcomes and risks) Then keep meetings only for:\nhigh-stakes disagreements fast synthesis when writing would be slower sensitive topics where nuance matters What changes when you apply this When artifacts exist:\nfewer people need to attend fewer meetings are needed to “remember” decisions stick because they’re recorded newcomers can catch up without draining the room The calendar becomes a tool again, not a symptom.\nConcrete example A product team had three recurring meetings: backlog grooming, roadmap alignment, stakeholder sync.\nThey introduced two artifacts:\na one-page Work Brief for each initiative a Decision Log updated weekly Result:\nstakeholder sync became a written update + optional Q\u0026amp;A roadmap alignment became asynchronous comments on briefs only disagreements triggered a meeting They didn’t “fix meetings.” They made meetings unnecessary.\nA simple checklist / rule of thumb A simple checklist What artifact will exist after this meeting that didn’t exist before? Who is the Directly Responsible Individual (DRI), and what decision are they empowered to make? What would make this meeting obsolete next month? Can the first 80% be done in writing? Are we meeting because we don’t trust the system? Rule of thumb If a recurring meeting has no artifact output, it’s a process bug.\n","permalink":"https://borggrech.com/thoughts/meeting-hell-is-a-symptom/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eThe core problem or tension you’re addressing (grounded in reality)\u003c/li\u003e\n\u003cli\u003eWhy this problem keeps showing up (systemic, not personal)\u003c/li\u003e\n\u003cli\u003eThe common but flawed response most teams default to\u003c/li\u003e\n\u003cli\u003eThe better framing or operating principle\u003c/li\u003e\n\u003cli\u003eWhat changes when this principle is applied\u003c/li\u003e\n\u003cli\u003eExample: a concrete, real-world scenario\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"open-with-the-real-problem\"\u003eOpen with the real problem\u003c/h2\u003e\n\u003cp\u003eYou open your calendar and it feels like a hostile takeover.\nRecurring meetings have multiplied, and somehow you still don’t feel aligned.\u003c/p\u003e","title":"Meeting Hell Is a Symptom: How to Kill 30% of Your Calendar"},{"content":"Outline Task assignment breaks when tasks become cheap and abundant Work design means shaping the workflow: inputs, outputs, owners, review Define “units of work” and acceptance criteria Reduce handoff tax with better interfaces (writing, templates, checklists) Build feedback loops that improve the system over time Example: turning a chaotic request stream into a stable workflow Task assignment assumes scarcity of execution Traditional management often looks like task assignment: break down work, distribute tasks, track completion.\nThat model assumes execution is the scarce resource.\nAI changes that. Execution becomes cheaper in many domains. The scarce resource becomes clarity: what to do, in what order, with what standards, under what constraints.\nIf managers keep focusing on assignment, they’ll create busy teams that produce a lot without moving outcomes.\nThe new management skill is designing how work flows through the system.\nWork design: inputs, outputs, owners, review Work design is not a grand reorg. It’s operational architecture.\nFor a given workflow, a manager should be able to answer:\nWhat triggers the work (inputs)? What is the expected output? Who is accountable for the outcome? What standards define “good”? What review happens, and when? What happens when we’re unsure? These questions sound simple. They are often missing.\nWhen they’re missing, teams compensate with meetings, interruptions, and heroics. AI may speed up drafts, but the system remains chaotic.\nDesign the system and the team becomes calmer and faster.\nDefine the unit of work and acceptance criteria A workflow without a clear unit becomes noisy.\nFor example, “handle customer requests” is not a unit. “Resolve customer issue with documented outcome and next step” is.\nOnce the unit is clear, define acceptance criteria: correctness, completeness, risk boundaries, and traceability where needed.\nAI can help produce drafts that meet criteria, but only if the criteria exist.\nThis is where managers add leverage: making standards explicit so the team can operate with autonomy.\nReduce handoff tax with better interfaces Most organizational waste lives in handoffs.\nOne person does work, another inherits it without context, and time is lost reconstructing intent.\nWork design reduces handoff tax by improving interfaces:\none-pagers for decisions templates for repeated outputs checklists for quality-critical steps clear ownership fields (“who owns next?”) AI helps by accelerating these artifacts. But the manager’s job is to standardize them and make them normal.\nA team with good interfaces can scale without increasing meeting load.\nBuild feedback loops that improve the system Good work design includes learning.\nIf the workflow produces repeated errors, the fix shouldn’t be “try harder.” It should be: update the checklist, clarify the criteria, improve the template, or adjust decision rights.\nManagers who build feedback loops create compounding improvements.\nThis also keeps morale high. Competent people hate repeating preventable mistakes. When the system improves, people feel respected.\nAI can accelerate learning by summarizing patterns: common failure modes, frequent exceptions, recurring customer pain points. Use that signal to update the system.\nConcrete example: stabilizing a chaotic request stream Imagine an internal team flooded with requests: “Can you pull this report?” “Can you fix this config?” “Can you help with onboarding?”\nThey respond reactively. Work is constantly interrupted. Nothing feels finished.\nA manager redesigns the workflow:\nAll requests enter through a single intake form. Each request must state desired outcome and deadline. Requests are triaged weekly into categories with clear owners. The unit of work becomes “request resolved with outcome + notes.” High-risk requests require review; low-risk are handled quickly. Recurring requests become templates or self-serve docs. Within weeks, interruptions drop. Predictability increases. The team ships fewer things—but delivers more outcomes.\nThis is management as work design.\nA simple checklist Are we managing by assigning tasks, or by designing workflows? Do we have a clear unit of work for each recurring workflow? Are acceptance criteria explicit and shared? Are handoffs structured with templates, checklists, and written context? Do we have feedback loops to improve the system over time? Where are meetings compensating for missing design? ","permalink":"https://borggrech.com/thoughts/manager-skill-designing-work/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eTask assignment breaks when tasks become cheap and abundant\u003c/li\u003e\n\u003cli\u003eWork design means shaping the workflow: inputs, outputs, owners, review\u003c/li\u003e\n\u003cli\u003eDefine “units of work” and acceptance criteria\u003c/li\u003e\n\u003cli\u003eReduce handoff tax with better interfaces (writing, templates, checklists)\u003c/li\u003e\n\u003cli\u003eBuild feedback loops that improve the system over time\u003c/li\u003e\n\u003cli\u003eExample: turning a chaotic request stream into a stable workflow\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"task-assignment-assumes-scarcity-of-execution\"\u003eTask assignment assumes scarcity of execution\u003c/h2\u003e\n\u003cp\u003eTraditional management often looks like task assignment: break down work, distribute tasks, track completion.\u003c/p\u003e","title":"The New Manager Skill: Designing Work, Not Assigning Tasks"},{"content":"Outline Templates reduce cognitive load and increase consistency Standard work is how you scale quality without micromanaging Templates should encode decision rights, criteria, and ownership AI makes templates more powerful (drafting + adaptation), not less necessary Keep templates lightweight and alive (review and iterate) Example: a decision template that improves execution across teams Templates are a way to package judgment People often resist templates because they fear rigidity.\nBut good templates don’t remove thinking. They remove repeated framing work so thinking can focus on the hard parts.\nIn a growing organization, inconsistency is expensive. Different people reinvent the same docs, decisions, and processes. Quality varies. Handoffs break.\nTemplates create a shared interface. They encode what matters. They reduce the chance that critical context is omitted.\nThis is strategy because it shapes how the organization behaves at scale.\nStandard work scales quality without adding managers Standard work sounds industrial. In knowledge work, it’s often the difference between a stable system and perpetual chaos.\nWhen you have standard work:\nnew hires ramp faster errors become easier to detect outcomes become more consistent review becomes lighter because expectations are clear Without it, quality depends on who is involved. That’s fragile.\nAI increases output speed, which increases the need for stable standards. Templates are one of the cheapest ways to provide that stability.\nTemplates should encode ownership and decision rights The most valuable templates do more than provide structure. They encode governance.\nA good decision template includes:\nthe decision to be made the decider the owner of execution the review date constraints and non-negotiables what would change the decision A good incident template includes:\nwhat happened customer impact owner root cause preventive change follow-up date When templates encode ownership, work stops floating. It lands.\nThis reduces politics and increases speed because authority is clear.\nAI makes templates more powerful, not obsolete Some assume AI makes templates unnecessary because you can “just ask the model.”\nThat misses the point.\nIf your template encodes your organization’s standards, the AI can draft within those standards. If you have no template, the AI will draft within generic internet patterns.\nTemplates give AI constraints, tone, and completeness.\nYou can also use AI to improve templates: identify missing fields, propose clearer language, detect recurring omissions, and generate examples.\nThe organization becomes more consistent because both humans and tools operate within the same structure.\nKeep templates alive: small review, regular iteration Templates fail when they become stale.\nIf people treat templates as “compliance,” they’ll fill them out mechanically or bypass them. If they treat templates as tools, they’ll improve them.\nA simple practice:\nreview templates quarterly ask: what fields are ignored? what fields are missing? update based on real failure modes and new needs Template evolution is how you keep standard work aligned with reality.\nA dead template is worse than no template. A living template is a compounding asset.\nConcrete example: a decision template that improves execution Imagine cross-team decisions constantly reappear: prioritization, process changes, customer commitments.\nYou introduce a decision template:\nContext (what’s happening, why now) Decision (the choice) Options (2–3 paths) Recommendation (with tradeoffs) Decider + date Owner + next actions Risks + mitigation Review date Within weeks, debates improve. People stop arguing over missing context. Execution improves because owners and next steps are explicit. Decisions stick because they’re recorded and reviewed.\nThis template becomes a quiet backbone. It scales management without creating more management.\nA simple checklist Where are we reinventing the same artifacts repeatedly? Which missing fields cause the most mistakes (owner, criteria, risks)? Can we encode decision rights and ownership into templates? Do templates make work easier—or feel like compliance? Do we review and update templates based on real failure modes? Are AI tools drafting within our templates and standards? ","permalink":"https://borggrech.com/thoughts/templates-as-strategy/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eTemplates reduce cognitive load and increase consistency\u003c/li\u003e\n\u003cli\u003eStandard work is how you scale quality without micromanaging\u003c/li\u003e\n\u003cli\u003eTemplates should encode decision rights, criteria, and ownership\u003c/li\u003e\n\u003cli\u003eAI makes templates more powerful (drafting + adaptation), not less necessary\u003c/li\u003e\n\u003cli\u003eKeep templates lightweight and alive (review and iterate)\u003c/li\u003e\n\u003cli\u003eExample: a decision template that improves execution across teams\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"templates-are-a-way-to-package-judgment\"\u003eTemplates are a way to package judgment\u003c/h2\u003e\n\u003cp\u003ePeople often resist templates because they fear rigidity.\u003c/p\u003e\n\u003cp\u003eBut good templates don’t remove thinking. They remove repeated framing work so thinking can focus on the hard parts.\u003c/p\u003e","title":"Templates as Strategy: The Quiet Power of Standard Work"},{"content":"Outline Start with a real workflow, not a tool (support, onboarding, planning) Define the outcome and the unit of work (“done”) Map the workflow: steps, handoffs, decisions, failure points Add AI where it reduces friction (drafting, triage, retrieval), not where it hides risk Add guardrails: data rules, risk tiers, review loops, escalation Example: redesigning quarterly planning to be writing-first and decision-driven Start with a workflow, not a tool The fastest way to waste time is to start with a tool and go hunting for a use case.\nInstead, start with a workflow that already matters: something that consumes time, creates friction, or produces recurring mistakes.\nExamples:\nsupport triage and resolution onboarding and time-to-value incident response planning and reporting proposal generation and deal qualification Workflows are where value is created. Tools are just leverage.\nIf you redesign the workflow, AI becomes a multiplier. If you don’t, AI becomes a faster way to stay stuck.\nDefine the outcome and the unit of work Before you touch the workflow, define two things:\nOutcome: what changes in the world when this workflow works.\nUnit of work: the smallest chunk you can ship and review that produces the outcome.\nIf you can’t define these, you can’t measure success. You also can’t govern risk.\nAI will tempt you to optimize for artifact production: more summaries, more drafts, more tickets. That’s not the outcome.\nAnchor the redesign in what “done” actually means.\nMap the workflow: steps, handoffs, decisions, failure points Now map the current workflow. Keep it simple:\nTrigger: what starts the work? Steps: what happens, in what order? Handoffs: where does context transfer? Decisions: what choices are made, by whom? Failure points: where do mistakes or delays occur? The goal is not perfect documentation. It’s identifying leverage points.\nMost workflow waste is in handoffs and unclear decisions.\nAI can help analyze logs, tickets, and notes to identify patterns. But the map should be understandable to the humans doing the work.\nIf people can’t recognize their workflow in your map, it won’t stick.\nAdd AI where it reduces friction without hiding risk AI is strongest in a few areas:\ndrafting first versions summarizing and structuring messy input retrieval from internal knowledge triage and classification generating checklists and options AI is weaker where authority and truth are critical:\npromises to customers financial commitments security claims irreversible production changes This doesn’t mean “never use AI.” It means place it where it helps and pair it with verification where it matters.\nYour design should include explicit “verify mode” steps for high-risk outputs.\nAdd guardrails: data rules, risk tiers, review, escalation Now add governance that helps:\nData rules: what can be used, where Risk tiers: internal draft vs external promise vs execution-impacting Review loops: sampling review to catch drift early Escalation path: what happens when unsure Guardrails are not a policy document. They are workflow steps.\nIf guardrails are not embedded in the workflow, they won’t be followed.\nThe goal is speed with trust. Guardrails are how you get both.\nConcrete example: redesigning quarterly planning Imagine quarterly planning is meeting-heavy, vague, and late. People bring slides. Decisions drift. Execution suffers.\nRedesign using the framework:\nOutcome: “A small set of priorities with owners, constraints, and review cadence.” Unit of work: “Decision-ready one-pager per initiative.” Workflow map: intake → proposal docs → review → decisions → published plan → monthly review. AI placement: draft one-pagers from notes, summarize input, identify dependencies, propose risks. Guardrails: decider named, risk tier for public commitments, review loop for plan quality, escalation path for cross-team conflicts. Result: fewer meetings, clearer decisions, and a plan people can execute. AI accelerates drafting and synthesis, but the governance holds the bar.\nA simple checklist Which workflow matters enough to redesign first? What is the outcome, and what is the unit of work? Where are the handoffs, unclear decisions, and repeat failure points? Where can AI reduce friction (drafting, triage, retrieval) safely? What guardrails are needed (data rules, risk tiers, review, escalation)? How will we detect drift early and update the system? ","permalink":"https://borggrech.com/thoughts/framework-redesigning-workflows-ai/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eStart with a real workflow, not a tool (support, onboarding, planning)\u003c/li\u003e\n\u003cli\u003eDefine the outcome and the unit of work (“done”)\u003c/li\u003e\n\u003cli\u003eMap the workflow: steps, handoffs, decisions, failure points\u003c/li\u003e\n\u003cli\u003eAdd AI where it reduces friction (drafting, triage, retrieval), not where it hides risk\u003c/li\u003e\n\u003cli\u003eAdd guardrails: data rules, risk tiers, review loops, escalation\u003c/li\u003e\n\u003cli\u003eExample: redesigning quarterly planning to be writing-first and decision-driven\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"start-with-a-workflow-not-a-tool\"\u003eStart with a workflow, not a tool\u003c/h2\u003e\n\u003cp\u003eThe fastest way to waste time is to start with a tool and go hunting for a use case.\u003c/p\u003e","title":"A Practical Framework for Redesigning Workflows With AI"},{"content":"Outline “Good” must be defined at the outcome level, not the artifact level Use acceptance criteria: correctness, completeness, tone, and risk Provide reference examples (good/bad) to reduce ambiguity Separate drafting from verifying (different mental modes) Sampling review is the fastest way to keep standards stable Example: customer email drafts that are fast—but safely correct If “good” is implicit, AI will expose the gap Before AI, many teams relied on tacit standards. A senior person “just knew” what a good response looked like. A strong engineer “just knew” how careful to be.\nAI changes that. When output is cheap, the variance increases. You’ll see more borderline work, more plausible-sounding mistakes, more inconsistency.\nThis isn’t because people became worse. It’s because the system now produces more surface area.\nIf you want standards to hold, you need to make “good” explicit.\nNot as a lecture. As a tool people can use.\nDefine good at the outcome level A common mistake is defining good as “a well-written email” or “a complete document.”\nThose are artifacts. The outcome is what the reader can do with it.\nFor a support response, the outcome might be “customer can resolve the issue without a follow-up.” For a proposal, “customer can make a decision.” For internal analysis, “leaders can act without guessing.”\nAI can generate artifacts that look polished while failing the outcome.\nSo define good like this: “After reading this, a reasonable person should be able to do X, with Y constraints, without Z risk.”\nThat’s a standard you can audit.\nUse acceptance criteria: correctness, completeness, tone, risk In practice, “good” becomes a short set of criteria.\nA reliable set often includes:\nCorrectness: claims are true, verifiable, and not overstated Completeness: includes the necessary steps, caveats, and next actions Tone: calm, respectful, non-defensive, not overly confident Risk: avoids promises, legal claims, or security assertions without approval These criteria are boring. That’s the point. They create consistency.\nIf the criteria are unclear, people will optimize for what’s visible: eloquence. AI is great at eloquence. Eloquence is not the job.\nReference examples reduce ambiguity faster than rules Rules are abstract. Examples are concrete.\nCreate a small library:\ntwo examples of “excellent” two examples of “acceptable” two examples of “not acceptable,” with why This is especially effective for tone and for risk boundaries.\nPeople don’t want to guess. If you show them what you mean, the team converges quickly.\nAI can help generate initial examples, but the human job is selecting the ones that match your standards and context.\nDrafting and verifying are different mental modes One of the quiet dangers of AI is blending drafting and verification.\nWhen a tool produces a confident answer, your brain wants to move on. You read it as if it’s already true.\nStrong teams separate modes:\nDraft mode: generate options fast, explore phrasing, structure Verify mode: check claims, sources, assumptions, edge cases You can even assign different people or different moments for these modes.\nThis isn’t paranoia. It’s process design.\nIf you collapse the modes, you ship faster—until you ship a mistake that costs you trust.\nSampling review keeps the bar stable without slowing the team You don’t need to review every output. You need to review enough to detect drift.\nA good cadence:\nweekly sampling of customer-facing outputs a clear rubric based on your acceptance criteria feedback that updates the examples and criteria This creates a learning loop. Over time, you need less review because the system stabilizes.\nWhen review is predictable and fair, people accept it. When review is random and punitive, they hide.\nConcrete example: AI-drafted customer emails Imagine customer success uses AI to draft emails about configuration changes. The drafts are fast and polite, but occasionally wrong: wrong feature name, wrong step order, subtle overpromises.\nYou define “good” with acceptance criteria:\nAny instruction must be verified in the product or docs. Any promise about timelines requires approval. Include one clear “next action” for the customer. You add examples: a good email, a bad email, and a revised version.\nThen you run sampling review: each week, review 10 emails across the team. You track recurring failure modes and fix the system: update internal docs, update prompts, and clarify boundaries.\nResult: speed stays. Trust improves. The team stops fearing AI and starts using it responsibly.\nA simple checklist Have we defined “good” as an outcome, not a polished artifact? Do we have acceptance criteria (correctness, completeness, tone, risk)? Do we have a small set of good/bad examples people can copy? Do we separate drafting from verifying in our workflow? Do we sample-review outputs and feed learnings back into the system? Are risk boundaries explicit (what requires approval, what doesn’t)? ","permalink":"https://borggrech.com/thoughts/defining-good-with-ai/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003e“Good” must be defined at the outcome level, not the artifact level\u003c/li\u003e\n\u003cli\u003eUse acceptance criteria: correctness, completeness, tone, and risk\u003c/li\u003e\n\u003cli\u003eProvide reference examples (good/bad) to reduce ambiguity\u003c/li\u003e\n\u003cli\u003eSeparate drafting from verifying (different mental modes)\u003c/li\u003e\n\u003cli\u003eSampling review is the fastest way to keep standards stable\u003c/li\u003e\n\u003cli\u003eExample: customer email drafts that are fast—but safely correct\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"if-good-is-implicit-ai-will-expose-the-gap\"\u003eIf “good” is implicit, AI will expose the gap\u003c/h2\u003e\n\u003cp\u003eBefore AI, many teams relied on tacit standards. A senior person “just knew” what a good response looked like. A strong engineer “just knew” how careful to be.\u003c/p\u003e","title":"What \"Good\" Looks Like When AI Touches the Output"},{"content":"Outline Services teaches you what customers actually pay for, if you listen\nThe trap: custom work that looks like product progress\n“No” is a product capability: clear boundaries, clear alternatives\nStandardize outcomes, not requests (same value, fewer variants)\nSales and delivery must share one definition of “supported”\nExample: turning repeated delivery work into a product workflow\nServices is the fastest way to learn, if you extract the signal Services work puts you close to reality. You see how customers behave, what they struggle with, and what they value enough to fund.\nThat’s a huge advantage.\nBut services also rewards responsiveness. If you can say “yes” quickly, you win deals. If you can bend the system, you keep customers happy in the short term.\nProduct requires a different discipline: repeating the same solution across many customers. That repetition is what creates leverage.\nThe transition succeeds when you keep the learning from services and remove the dependency on custom delivery.\nThe trap: custom work that looks like product progress The most dangerous phase is when you think you’re building product, but you’re still selling custom work.\nIt often looks like this:\nA customer asks for “one small change.”\nThe team ships it quickly.\nSales sells another deal based on that change.\nSoon you have five versions of the same workflow.\nSupport becomes a memory game.\nYou can even call it a “roadmap.” But it’s not a roadmap. It’s a collection of exceptions.\nAI can accelerate this trap. It makes it easier to produce variant: different prompts, different scripts, different integrations, without feeling the cost immediately. The cost shows up later in confusion and maintenance.\nThe test is simple: can you deliver the value without rethinking the solution every time?\n“No” is a product capability Many founders and operators avoid saying no because it feels like losing revenue. In services, that’s often true.\nIn product, the cost of “yes” is compounding. Each exception creates future work: docs, support, testing, migration, and mental load.\nSaying no doesn’t mean being rigid. It means being explicit:\nwhat you support\nwhat you don’t support\nwhy\nwhat the alternative is\nwhat would need to be true to change the boundary\nThis is not sales theatre. It’s operational integrity.\nCustomers can accept boundaries if you communicate them clearly and early. What they won’t accept is surprise.\nStandardize outcomes, not requests One useful shift is to standardize around outcomes rather than features.\nCustomers request features in their language: “We need this field,” “We need that report,” “We need this workflow.” Those requests are often proxies for an outcome: “We need fewer errors,” “We need faster approvals,” “We need auditability.”\nWhen you identify the outcome, you can offer a standard solution that achieves it without absorbing their entire internal complexity.\nAI can help here by summarizing patterns across customer conversations: repeated pain points, common language, recurring exceptions. But the human work is deciding what becomes standard and what stays out.\nThe product isn’t the sum of requests. It’s the curated set that stays coherent.\nSales and delivery must share one definition of “supported” In a services-led organization, sales can promise almost anything because delivery will “figure it out.”\nThat’s not evil. It’s how you survive early.\nAs you productize, that must change. Sales needs a clear definition of supported configurations, timelines, and outcomes. Delivery needs permission to push back when deals drift.\nThe practical mechanism is a short “supported” document: what we do, what we don’t, what requires a paid exception, and who approves exceptions.\nWhen sales and delivery disagree, the customer becomes the referee. That’s always bad.\nWith aligned boundaries, you build trust internally and externally.\nConcrete example: repeated onboarding work becomes a product workflow Imagine your team does onboarding for each customer. Every customer wants a slightly different setup, and the team obliges.\nOver time you notice a pattern: 80% of onboarding effort is the same. The differences are often cosmetic or internal preference.\nYou decide to productize onboarding:\nA standard intake form replaces long email threads.\nA standard configuration workflow replaces ad hoc scripts.\nA standard “first value” milestone replaces “onboarding complete.”\nWhen a customer requests a unique setup, you respond with a clear boundary: “We support A, B, C. If you need D, here’s the cost and the tradeoff.”\nAt first, it feels uncomfortable. Then something happens: delivery gets faster, support gets simpler, and the product becomes easier to explain.\nThe team stops being heroic and starts being reliable.\nA simple checklist What are the 10 most common customer requests, and what outcomes do they represent?\nWhich “small” customizations are actually long-term maintenance liabilities?\nDo we have a written definition of “supported,” owned by product + delivery?\nAre exception approvals explicit (who can say yes, and at what cost)?\nCan we standardize on outcomes while offering configurable paths?\nDo sales and delivery share the same boundaries in customer conversations?\nAre we building a product, or building many one-off versions of the same thing?\n","permalink":"https://borggrech.com/thoughts/services-to-product-saying-no/","summary":"\u003ch2 id=\"outline\"\u003eOutline\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eServices teaches you what customers actually pay for, if you listen\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eThe trap: custom work that looks like product progress\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e“No” is a product capability: clear boundaries, clear alternatives\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eStandardize outcomes, not requests (same value, fewer variants)\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eSales and delivery must share one definition of “supported”\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eExample: turning repeated delivery work into a product workflow\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch2 id=\"services-is-the-fastest-way-to-learn-if-you-extract-the-signal\"\u003eServices is the fastest way to learn, if you extract the signal\u003c/h2\u003e\n\u003cp\u003eServices work puts you close to reality. You see how customers behave, what they struggle with, and what they value enough to fund.\u003c/p\u003e","title":"From Services to Product: The Hard Part Is Saying No"},{"content":"\nI’m Edward Borg Grech, founder and CEO of Bluefort, based in Malta.\nMy career began in software development. Over time, I moved from writing code into the messier and more interesting work of building a company: delivering for customers, shaping products, hiring people, selling, making trade-offs and taking responsibility for outcomes.\nFrom software to company-building I co-founded Bluefort and have spent years helping it evolve from a services-led software business into a company with its own enterprise products and recurring-revenue ambitions.\nThat journey has given me an unusually broad operating perspective. I have seen the same problem from different seats: as a developer trying to make the system work, as a delivery leader trying to make the customer successful, as a product builder deciding what not to build, and as a CEO thinking about markets, people, capital and long-term direction.\nIt has also made me sceptical of elegant theories that ignore implementation. Technology matters. But ownership, incentives, distribution, judgment and execution usually determine whether the technology creates lasting value.\nWhy I’m interested in AI now AI is becoming more than another productivity tool. Systems are moving from generating content towards planning, deciding and acting. That changes the organisational question.\nThe interesting problem is no longer simply, “Where can we automate?” It is also:\nWho remains accountable when a system acts? How do we verify outcomes without recreating all the work we automated? Which decisions should remain human, and why? How do organisations change when execution becomes abundant? What happens to agency, dignity and meaning when machines can do more of what previously made people economically valuable? I am particularly interested in the idea of accountable autonomy: how we gain the benefits of increasingly capable systems without allowing responsibility to become vague or disappear altogether.\nMalta, Europe and the wider transition I live and work in Malta. I see that as an advantage rather than a limitation.\nSmall countries can be unusually good places to test ideas: institutions are closer together, networks are denser and the distance between an idea and a decision can be shorter. At the same time, Malta sits inside the European Union and therefore inside one of the most consequential debates about how AI should be built, governed and deployed.\nI’m interested in what Europe needs to build, not only what it chooses to regulate — and in how smaller countries can participate meaningfully in that future.\nBackground My formal background spans Business Information Systems, an MSc in Software Engineering and an MBA from Henley Business School.\nThe more useful education, however, has been building through real constraints: customers who need outcomes, products that have to scale, teams that need clarity, and decisions where there is rarely perfect information.\nI write here to make my thinking more precise and to find people working seriously on similar questions.\nRead my writing · What I’m focused on now · Email · LinkedIn ","permalink":"https://borggrech.com/about/","summary":"\u003cp\u003e\u003cimg alt=\"Borg Grech\" loading=\"lazy\" src=\"/images/headshot.jpg\"\u003e\u003c/p\u003e\n\u003cp\u003eI’m Edward Borg Grech, founder and CEO of \u003ca href=\"https://www.bluefort.eu/\"\u003eBluefort\u003c/a\u003e, based in Malta.\u003c/p\u003e\n\u003cp\u003eMy career began in software development. Over time, I moved from writing code into the messier and more interesting work of building a company: delivering for customers, shaping products, hiring people, selling, making trade-offs and taking responsibility for outcomes.\u003c/p\u003e\n\u003ch2 id=\"from-software-to-company-building\"\u003eFrom software to company-building\u003c/h2\u003e\n\u003cp\u003eI co-founded Bluefort and have spent years helping it evolve from a services-led software business into a company with its own enterprise products and recurring-revenue ambitions.\u003c/p\u003e","title":"About"},{"content":"Updated July 2026.\nThis page is a snapshot of what I’m spending my time thinking about and working on now. It will change as the work changes.\nOperating I continue to lead Bluefort as founder and CEO, with a focus on building a durable enterprise software business: strengthening products, commercialisation, partnerships and the systems that allow a company to scale without becoming slower than the problems it exists to solve.\nI remain especially interested in the difficult boundary between services and product — how to learn deeply from customers without allowing every customer request to become the roadmap.\nExploring I’m spending increasing time on what happens as AI moves from assistance towards autonomy.\nThe questions that interest me most sit between technology and organisational design: agentic systems, verification, accountability, decision rights, auditability and the new operating models that become possible when more execution can be delegated to machines.\nI’m also exploring what kinds of new companies become possible in this environment, particularly businesses that combine deep implementation with reusable intellectual property rather than choosing between pure consulting and pure software.\nThinking about Europe and Malta I’m interested in whether Europe can turn its strengths — talent, industrial depth, trusted institutions and a large market — into genuine technological capability rather than assuming that good regulation alone will create competitiveness.\nFrom Malta, I’m particularly curious about the role small countries can play as focused environments for experimentation, applied research and cross-sector collaboration.\nWriting My writing is becoming broader than the future of work alone. I’m increasingly interested in AI and institutions, accountable autonomy, abundance, business-building, judgment, human agency and the choices Europe will make during this transition.\nThe aim is not to publish constantly. It is to publish when I have something worth making more precise.\nRead the latest writing · About me · Email ","permalink":"https://borggrech.com/now/","summary":"\u003cp\u003e\u003cem\u003eUpdated July 2026.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis page is a snapshot of what I’m spending my time thinking about and working on now. It will change as the work changes.\u003c/p\u003e\n\u003ch2 id=\"operating\"\u003eOperating\u003c/h2\u003e\n\u003cp\u003eI continue to lead \u003ca href=\"https://www.bluefort.eu/\"\u003eBluefort\u003c/a\u003e as founder and CEO, with a focus on building a durable enterprise software business: strengthening products, commercialisation, partnerships and the systems that allow a company to scale without becoming slower than the problems it exists to solve.\u003c/p\u003e\n\u003cp\u003eI remain especially interested in the difficult boundary between services and product — how to learn deeply from customers without allowing every customer request to become the roadmap.\u003c/p\u003e","title":"Now"}]