Return-Path: Delivered-To: alvarochf@traceseis.com Received: from cs510.bluehost.com by cs510.bluehost.com with LMTP id cB80BRBhJWqBXAgAWiBQxw (envelope-from ) for ; Sun, 07 Jun 2026 06:16:16 -0600 Return-path: Envelope-to: alvarochf@traceseis.com Delivery-date: Sun, 07 Jun 2026 06:16:16 -0600 Received: from v555.v5375b7fa.use4.send.mailgun.net ([159.112.244.55]:47708) by cs510.bluehost.com with esmtps (TLS1.2) tls TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 (Exim 4.99.2) (envelope-from ) id 1wWCQJ-00000002JWn-0Eu1 for alvarochf@traceseis.com; Sun, 07 Jun 2026 06:16:16 -0600 DKIM-Signature: a=rsa-sha256; v=1; c=relaxed/relaxed; d=mg2.substack.com; q=dns/txt; s=mailo; t=1780834558; x=1780841758; h=List-Unsubscribe-Post: List-Unsubscribe: List-Post: List-Id: List-Archive: List-Owner: Reply-To: In-Reply-To: References: Sender: Sender: Date: Message-Id: To: To: From: From: Subject: Subject: Content-Type: Mime-Version; bh=/tvcTzUfCRP6tW1F7eUtoaUaH+MxpE2Gz6N64KuJitw=; b=HZS53OQh1pekMNLoOfenSeWHAD645CVl+V67hURpWifdFjCuCiY4yOuCiamSJlpJ1vYCdmXLAZxwtOJO1YfPB7i8QqEhgxvO6MNK+nysHiaAmJT1p0zlpDw9yM/4AZbhVjWEDQ7VjEPDCw893dp+rMMEg2CswdVZVvnlfY4gbOk= X-Mailgun-Sid: WyI3MDdlYyIsImFsdmFyb2NoZkB0cmFjZXNlaXMuY29tIiwiNGQxODU4Il0= Received: by bff2046af41f4c90e7cdb8dac6f59ae17c58da5745cd25c7cb268e02b8990af1 with HTTP id 6a2560feaaec93d28b7088b2; Sun, 07 Jun 2026 12:15:56 GMT X-Mailgun-Sending-Ip: 159.112.244.55 X-Mailgun-Batch-Id: 6a2560fcd26877452428a660 Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="25e5463c370c0f1c85cf24a5210d3e83d9aaba1dd37ca78e27de70df071a" Subject: Execution Is Getting Cheap. Ambition Is Getting Weird. From: Zach Chmael from Don't Feed The Algorithm To: alvarochf@traceseis.com X-Mailgun-Tag: post X-Mailgun-Track-Clicks: false Message-Id: <20260607121444.3.751a365bd5b4b841@mg2.substack.com> Date: Sun, 7 Jun 2026 12:14:44 +0000 Feedback-Id: post-199523246:cat-post:pub-4987907:substack Sender: Zach Chmael from Don't Feed The Algorithm References: In-Reply-To: Reply-To: Zach Chmael from Don't Feed The Algorithm List-Owner: List-Url: List-Archive: List-Id: List-Post: List-Unsubscribe: List-Unsubscribe-Post: List-Unsubscribe=One-Click X-Mailgun-Variables: {"category": "post", "email_generated_at": "1780834551340", "is_freemail": "true", "post_audience": "everyone", "post_id": "199523246", "post_type": "newsletter", "pub_community_enabled": "true", "publication_id": "4987907", "subdomain": "zchmael", "user_id": "347258985"} X-Spam-Status: No, score=-0.8 X-Spam-Score: -7 X-Spam-Bar: / X-Spam-Flag: NO --25e5463c370c0f1c85cf24a5210d3e83d9aaba1dd37ca78e27de70df071a Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable View this post on the web at https://zchmael.substack.com/p/execution-is-ge= tting-cheap-ambition For a long time, execution was a very convenient excuse. Not a fake excuse. A real one. Making things was hard, man. Building a comp= any was harder. Publishing consistently was hardest. Designing a product, l= aunching a campaign, writing the essay, editing the video, building the lan= ding page, analyzing the data, making the deck, sending the email, followin= g up, doing the second version, doing the third version after the second ve= rsion revealed you were maybe not the visionary you briefly became in the s= hower. Hard. Annoying. Expensive. Time-consuming in the specific way that makes you look up at 7:48 p.m. and = realize you have eaten one protein bar and developed a personal relationshi= p with a spreadsheet. So we all got to say some version of the same thing=E2=80=A6 I would, but I do not have the time. I would, but I do not have the team. I would, but I do not have the budget. I would, but I do not have the design skills, the writing skills, the resea= rch help, the analytics setup, the editor, the operator, the intern, the ag= ency, the growth person, the mysterious adult in the room who apparently kn= ows how to make a go-to-market motion stop looking like a Google Doc fell d= own the stairs. Some of that was true. A lot of it still is. But AI is starting to make those excuses less comfortable. Not gone. Please= do not become the person who says =E2=80=9Cyou can build anything now=E2= =80=9D after prompting a chatbot to make a todo app that stores everyone=E2= =80=99s passwords in a text file named `final_final_secure.txt`. The world = is still full of real constraints. Distribution is hard. Quality is hard. T= rust is hard. Money remains, inconveniently, money. Still, something has shifted. The gap between idea and artifact is shrinking. You can get a draft faster.= A prototype faster. A visual direction faster. A research summary faster. = A first campaign faster. A rough analysis faster. A weird little workflow t= hat does the annoying first pass faster. The blank page is less blank. The = first version is less expensive. The distance between =E2=80=9CI wonder if= =E2=80=9D and =E2=80=9Chere is something I can react to=E2=80=9D is collaps= ing. Which sounds like pure good news. It is good news. It is also going to make ambition extremely weird. when execution gets cheaper, taste gets louder The first obvious shift is that taste becomes more valuable. I have been be= ating this drum enough that the drum has probably filed a formal complaint,= but it remains true all the same=E2=80=A6 when everyone can make more stuf= f, the advantage moves to knowing what should exist. But there is another layer underneath taste. Taste is not just the ability to say =E2=80=9Cthis is good=E2=80=9D or =E2= =80=9Cthis is bad.=E2=80=9D Taste is a relationship with desire. It is know= ing what you are trying to make real. It is having enough internal signal t= o recognize when the machine produced something technically correct and spi= ritually unemployed. It is being able to look at ten decent options and fee= l the one that has a pulse. That sounds nice until you realize how much of modern work has trained us t= o avoid that feeling. A lot of work is built around external validation. What is the benchmark? W= hat is the competitor doing? What does the algorithm reward? What did the c= ustomer ask for? What did the board like? What did the previous campaign pr= ove? What does the dashboard say? What would be safest to ship before the m= eeting where everyone pretends the problem is prioritization and not fear h= olding a reusable water bottle it received at a SaaS offsite? These are not bad questions. They are necessary questions. But they are not= enough when execution gets cheap. Because if you can make ten versions, the question becomes which version do= you actually believe in? If you can launch three tests, what are you trying to learn? If you can generate fifty angles, which angle deserves your name on it? If you can build the prototype over a weekend, what is the product trying t= o make possible? AI does not remove the need for ambition. It removes some of the camouflage= around the lack of it. Rude technology, honestly. the comfort of being blocked There is a weird comfort in being blocked by logistics. If the thing is impossible because you do not have the resources, you never= have to find out if the idea is good. If the project requires a team you d= o not have, you never have to discover whether you were willing to lead it.= If the essay would take two weeks and you only have two hours, you can kee= p the essay in the pristine little museum of things you would totally write= if capitalism stopped emailing you. I am not judging this. I have lived here. I have paid rent in this neighbor= hood. I have furnished the apartment with elaborate notes apps and the emot= ional support tabs of half-finished ideas. Constraints protect us from exposure. They let the dream stay perfect becau= se it never has to enter the world and become a version one with typos, unc= lear positioning, and one guy in the comments explaining that technically t= he Romans had content engines. AI messes with this bargain. It says: okay, maybe you cannot finish it perfectly today. But you can make= a first version. You can test the shape. You can draft the landing page. Y= ou can mock the product. You can turn the idea into something visible enoug= h to be judged. Terrible. Useful. Terrible because now the idea has a body, and bodies are vulnerable. Useful= because this is where real work starts. The dream does not become serious when you imagine it harder. It becomes se= rious when it has to survive contact with reality. AI makes contact cheaper. That means more people are going to have to answer a question they have bee= n outsourcing to friction=E2=80=A6 Do I actually want this? productivity is not the same as direction This is the part that makes me nervous about a lot of AI productivity cultu= re. The pitch is usually more output. More speed. More force. More capacity. Mo= re drafts, more tests, more campaigns, more clips, more analysis, more auto= mation, more personal operating systems, more little agents running around = your digital life like caffeinated interns with root access. Cool. But more output in the wrong direction is not progress. It is a Peloton for= your anxiety. You are moving. The numbers are changing. There is sweat. A = screen is congratulating you. You have not gone anywhere. I love systems. I love workflows. I love the idea that a person with good t= aste and a small stack of agents can do work that used to require a whole d= epartment and a conference room full of snacks nobody admits to liking. Thi= s is obviously part of what we are building toward at Averi [ https://link.= sbstck.com/redirect/4d3aadca-375d-4d98-9454-6e7a2199f192?j=3DeyJ1IjoiNXFxeX= F4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ]=E2=80=A6 marketing syst= ems that remember, learn, draft, analyze, and help teams move faster withou= t turning everything into gray paste. But speed only helps if you know where you are going. Otherwise AI becomes a very powerful way to avoid the more embarrassing que= stion. Not =E2=80=9Chow do I make more?=E2=80=9D but =E2=80=9Cwhat am I try= ing to become by making all of this?=E2=80=9D That question is gross. It sounds like someone put a brand strategist and a= youth pastor in the same Patagonia store. I know. I hate it too. But it is there. For companies, it sounds like, what category are we trying to define? What = belief are we trying to change? What would we still refuse to say even if i= t got clicks? What would make this company matter if the market did not han= d us relevance for free? For people, it sounds like, what do I want to be known for? What work feels= alive enough to keep doing when the dopamine leaves? What am I willing to = practice in public? What am I no longer allowed to pretend I am waiting for= permission to start? AI makes the operational questions easier. It makes the existential ones louder. Sorry. the ambition gap I think we are going to see a strange split. Some people and companies will use AI to produce more of what they were alr= eady producing. More posts. More campaigns. More dashboards. More polite de= cks with the same old strategy wearing a new font. More activity. More moti= on. More artifacts. Others will use AI to raise the level of what they attempt. That is the ambition gap. The first group asks, =E2=80=9CHow can we do the same thing faster?=E2=80= =9D The second asks, =E2=80=9CWhat can we finally attempt because the first= 30 percent is no longer such a tax?=E2=80=9D That difference matters. If AI helps you publish five mediocre posts instead of one mediocre post, c= ongratulations, you have built a small weather system of mediocrity. If AI = helps you run the research, pressure-test the angle, create the draft, gene= rate variants, analyze the response, and tighten your judgment until the wo= rk gets sharper every week, different game. Same tools. Different ambition. If AI helps you make ten landing pages because you have ten segments and no= strategy, enjoy your content cul-de-sac. If AI helps you discover which se= gment is actually resonating, what language they use, what objection keeps = appearing, and what promise you can credibly own, now the machine is helpin= g you think. Same output category. Different job. The people who win will not be the ones who automate the most. They will be= the ones who use automation to attempt something they were previously too = constrained, too slow, or too scattered to attempt. That might be a company building a real content intelligence system instead= of a content calendar with seasonal depression. It might be a founder fina= lly turning scattered notes into a point of view. It might be a creator mov= ing from =E2=80=9CI post when inspired=E2=80=9D to a serious practice. It m= ight be a tiny team operating with the research discipline of a much bigger= one. The size of the team matters less than the size of the attempt. AI makes seriousness visible There is a thing I keep noticing in myself and in other people: AI exposes = seriousness. Not talent. Not intelligence. Seriousness. If someone is serious about writing, AI becomes material. They use it to ar= gue with themselves, find holes, test structures, get unstuck, produce clay= =2E Then they edit. They choose. They c= ut. They make the thing more theirs. If someone is not serious, AI becomes a costume. They ask for the finished = thing, skim it, maybe change two words, and ship something that sounds like= it was assembled during a layover by a committee of autocomplete ghosts. If a team is serious about strategy, AI becomes an amplifier. It helps them= collect customer language, test messages, build briefs, review work agains= t principles, and move faster with more memory. If a team is not serious, A= I becomes a fog machine for the same confusion. More words around the same = unresolved decisions. This is why I do not buy the idea that AI automatically makes everyone crea= tive or strategic or dangerous. It makes serious people more dangerous. It = makes unserious people louder. That is not a small distinction. The tool does not decide whether you care. It just makes the caring, or lack of caring, easier to see. effort is changing shape A common mistake is thinking cheap execution means effort stops mattering. No. Effort just moves. Less effort on the blank page. More effort on direction. Less effort on first drafts. More effort on selection. Less effort on mechanical production. More effort on taste, distribution, f= eedback, iteration, and the emotional horror of looking at your own work ho= nestly. This is good. I think. Mostly. Ask me again after I have reviewed 47 AI-ass= isted drafts from people who think =E2=80=9Cmake it more premium=E2=80=9D i= s feedback. The old effort was often physical or logistical. Can I make the thing? Can = I find the time? Can I learn the tool? Can I get someone to help? The new e= ffort is more psychological and editorial. Can I decide what matters? Can I= reject the easy version? Can I keep going after the first pass looks good = enough? Can I tolerate the gap between the artifact and the thing I meant? That last one is brutal. AI shrinks the gap between intention and artifact, but it does not close th= e gap between artifact and taste. In some ways, it makes that gap more anno= ying because now you can see so many near-misses so quickly. The machine ke= eps handing you versions that are almost there. Almost there is seductive. = Almost there whispers, =E2=80=9CShip me, coward.=E2=80=9D Do not trust almost there. Almost there is where good work goes to get a LinkedIn carousel and a quiet= life. The new effort is refusing almost there when almost there would be enough. permission is also getting cheaper There is another weird thing happening. AI does not only reduce execution cost. It reduces permission cost. You used to need someone to validate that you were allowed to try certain t= hings. A designer. A developer. An editor. A strategist. A person with a ti= tle that made the thing feel official. This was sometimes good because expe= rtise matters. It was sometimes bad because the gate was mostly vibes and c= alendar availability. Now you can make a version before anyone gives you permission. Not the fina= l version. Not always a good version. But a version. Enough to show, test, = learn, embarrass yourself in a contained environment, and improve. That changes the psychology of starting. The sentence =E2=80=9CI do not know how=E2=80=9D used to end the conversati= on. Now it starts a workflow. I do not know how to build this prototype. Okay, ask for a scaffold. I do not know how to structure this essay. Okay, make three outlines and fi= ght with them. I do not know how to analyze this data. Okay, get the first pass, then veri= fy it like an adult with trust issues. I do not know how to make this campaign feel less generic. Okay, generate o= ptions, compare them against your enemy list, and find where the real tensi= on is hiding. This is not a replacement for expertise. It is a way to begin before expert= ise fully arrives. And beginning before you feel ready is, unfortunately, where a lot of inter= esting work lives. the downside: everyone can perform ambition now Of course there is a dark version. If execution gets cheap, performance gets cheap too. People can perform ser= iousness at scale. They can create the artifacts of ambition without the su= bstance: the pitch deck, the manifesto, the content engine, the personal br= and, the founder thread, the AI-generated product mockup, the strategy memo= that looks expensive and says almost nothing. This is already everywhere. The internet is filling with people who have the output surface area of a m= edia company and the inner life of a vending machine. That sounds mean. It is mean. It is also sometimes true. The artifacts of ambition are no longer proof of ambition. A deck is not pr= oof. A landing page is not proof. A content calendar is not proof. A polish= ed essay is not proof. AI can help make all of those before lunch, especial= ly if lunch is one of those $19 salads that tastes like someone described n= utrition to a hedge fund. So what is proof? Consistency is proof. Iteration is proof. Taste under pressure is proof. The willingness to be specific is proof. The ability to learn from the market without becoming whatever the market m= ost recently rewarded is proof. The willingness to kill your own fine work because it is not the work is pr= oof. The artifact matters. But the pattern matters more. AI makes artifacts cheaper. Patterns are still expensive. what to do with this If you are building anything right now, I think the move is not just to ask= , =E2=80=9CHow can AI make this faster?=E2=80=9D Ask a better set of questions. What would I attempt if the first draft were free? What would I test if the= prototype took a weekend instead of a quarter? What would I publish if the= research assistant, editor, and repurposing machine were already in the ro= om? What would I build if I trusted myself to judge the outputs instead of = outsourcing the taste to the tool? Then ask the scarier version: What am I still not doing, even now that it is easier? That is the question with teeth. Because if the blocker was truly logistics, AI helps. If the blocker was fe= ar, identity, taste, or lack of desire, AI will politely generate a workaro= und while the real problem sits in the corner wearing noise-canceling headp= hones. This does not mean you need to become some maximalist productivity goblin. = Please do not turn your life into a dashboard where every sunrise is a KPI = and every friendship has a retention curve. The point is not to produce you= rself into dust. The point is to notice where the old constraints are loosening, and to rais= e the ambition instead of only raising the volume. Make fewer things if you need to. Make stranger things. Make sharper things. Make things that would have been too annoying to start before, too expensiv= e to explore, too technically irritating to test, too messy to explain in t= he meeting where everyone keeps saying =E2=80=9Cnet new=E2=80=9D like a pra= yer. Use the machine to buy back the courage to try. Not just the capacity to ship. the question I cannot stop thinking about A lot of people are going to use AI to become more efficient versions of wh= o they already were. That is fine. Useful, even. Many workflows deserve to be faster. Many tasks= deserve to be automated into the sun. Nobody needs to spiritually commune = with a CSV export at 11:34 p.m. unless that is your thing, in which case I = support your journey from a respectful distance. But the more interesting question is whether AI can help people and teams b= ecome more ambitious versions of themselves. Not louder. Not busier. Not more omnipresent on every channel until the audience starts developing = the digital equivalent of hives. More ambitious. More willing to attempt the thing they actually mean. More willing to build= the system, write the argument, define the category, challenge the default= , make the weird prototype, publish the sharper take, design the better rit= ual, refuse the safe version, and keep going after the machine has done the= easy part. That is the part I am watching. Because if execution gets cheap and ambition stays small, we get more stuff= =2E If execution gets cheap and ambition expands, we might get work that feels = impossible from the old world. Maybe that is naive. Probably a little. But I would rather be naive in that direction than spend the next decade op= timizing the production of things nobody wanted badly enough to make well. The machine can lower the floor. It cannot raise your ceiling for you. -zc P.S. If this made you realize your AI workflow is very good at helping you = avoid the actual question, welcome. Horrible club. Great snacks.=20 Subscribe for the rabbit hole for the overcaffeinated, where we keep asking= whether the machine made us more powerful or just gave our procrastination= a nicer UI. Rabbit hole for the overcaffeinated - Ethan Mollick on holding back the strange AI tide: https://www.oneusefult= hing.org/p/on-holding-back-the-strange-ai-tide [ https://link.sbstck.com/re= direct/562065af-ab56-4377-8870-d99d1da67d69?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbG= j64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] - Ethan Mollick on centaurs, cyborgs, and the jagged frontier of AI work: h= ttps://www.oneusefulthing.org/p/centaurs-and-cyborgs-on-the-jagged [ https:= //link.sbstck.com/redirect/6987f721-2644-41cd-b568-e5518f4515c4?j=3DeyJ1Ijo= iNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-HyoA ] - Andrej Karpathy=E2=80=99s =E2=80=9CSoftware Is Changing (Again)=E2=80=9D = talk, useful for thinking about what becomes easier when software becomes m= ore malleable: - Paul Graham=E2=80=99s =E2=80=9CHow to Do Great Work,=E2=80=9D still a use= ful gut check on ambition, taste, obsession, and seriousness: https://paulg= raham.com/greatwork.html [ https://link.sbstck.com/redirect/ec5117be-9b01-4= fd9-bdd4-601283e03f94?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8U= KGoFv6eWS-HyoA ] - Averi, where this shows up as the difference between making more marketin= g stuff and building a system that helps teams attempt better work: https:/= /www.averi.ai/ [ https://link.sbstck.com/redirect/557602da-a551-4ed9-9911-a= b03607410d0?j=3DeyJ1IjoiNXFxeXF4In0.h9dEDbGj64cTNudluPwbMYWklSu8UKGoFv6eWS-= HyoA ] Unsubscribe https://substack.com/redirect/2/eyJlIjoiaHR0cHM6Ly96Y2htYWVsLnN= 1YnN0YWNrLmNvbS9hY3Rpb24vZGlzYWJsZV9lbWFpbD90b2tlbj1leUoxYzJWeVgybGtJam96Tk= RjeU5UZzVPRFVzSW5CdmMzUmZhV1FpT2pFNU9UVXlNekkwTml3aWFXRjBJam94Tnpnd09ETTBOV= FV4TENKbGVIQWlPakU0TVRJek56QTFOVEVzSW1semN5STZJbkIxWWkwME9UZzNPVEEzSWl3aWMz= VmlJam9pWkdsellXSnNaVjlsYldGcGJDSjkudXlickdhWlpWeV83TThOY1JRYXExcjNRUmgtQ3d= XOFlIUEFnQlR5TU13YyIsInAiOjE5OTUyMzI0NiwicyI6NDk4NzkwNywiZiI6dHJ1ZSwidSI6Mz= Q3MjU4OTg1LCJpYXQiOjE3ODA4MzQ1NTEsImV4cCI6MjA5NjQxMDU1MSwiaXNzIjoicHViLTAiL= CJzdWIiOiJsaW5rLXJlZGlyZWN0In0.tQYBc0I0MWidp52mVQBSHX8Pwf_4v9yFILpD1q0T_Tw? --25e5463c370c0f1c85cf24a5210d3e83d9aaba1dd37ca78e27de70df071a Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable = Execution Is Getting Cheap. Ambition Is Getting Weird.
AI is making it easier to make the thing. Th= e stranger question is what happens when the bottleneck moves from effort t= o desire, from “can I do this?” to “what was I actually t= rying to build?"
͏     ­͏     ­&= #847;     ­͏     ­͏    = ; ­͏     ­͏     ­͏ &nbs= p;   ­͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­͏ =     ­͏     ­͏     = 3;͏     ­͏     ­͏   = 199; ­͏     ­͏     ­͏ &= nbsp;   ­͏     ­͏     ­= ;͏     ­͏     ­͏   Q= 99; ­͏     ­͏     ­͏ &n= bsp;   ­͏     ­͏     ­= ͏     ­͏     ­͏   ̳= 9; ­͏     ­͏     ­͏ &nb= sp;   ­͏     ­͏     ­&= #847;     ­͏     ­͏    = ; ­͏     ­͏     ­͏ &nbs= p;   ­͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­͏ =     ­͏     ­͏     = 3;͏     ­͏     ­͏   = 199; ­͏     ­͏     ­͏ &= nbsp;   ­͏     ­͏     ­= ;͏     ­͏     ­͏   Q= 99; ­͏     ­͏     ­͏ &n= bsp;   ­͏     ­͏     ­= ͏     ­͏     ­͏   ̳= 9; ­͏     ­͏     ­͏ &nb= sp;   ­͏     ­͏     ­&= #847;     ­͏     ­͏    = ; ­͏     ­͏     ­͏ &nbs= p;   ­͏     ­͏     ­&#= 847;     ­͏     ­͏    = ­͏     ­͏     ­͏  = ;   ­͏     ­͏     ­= 47;     ­͏     ­͏     = ­͏     ­͏     ­͏  =   ­͏     ­͏     ­T= 7;     ­͏     ­͏     &= #173;͏     ­͏     ­͏   =   ­͏     ­͏     ­͏= ;     ­͏     ­͏     &#= 173;͏     ­͏     ­͏   &= #8199; ­͏     ­͏     ­͏=     ­͏     ­͏     = 73;͏     ­͏     ­͏   &#= 8199; ­͏     ­͏     ­
<= table class=3D"email-body-container" role=3D"presentation" width=3D"100%" b= order=3D"0" cellspacing=3D"0" cellpadding=3D"0">
=
Forwarded this email? Subscribe here for more

Execution Is Getting Cheap. Am= bition Is Getting Weird.

AI is making it easier to make the thing. The = stranger question is what happens when the bottleneck moves from effort to = desire, from “can I do this?” to “what was I actually try= ing to build?"

= =
<= /table>
Zach Chmael
 
<= tr>
<= img class=3D"icon" src=3D"https://substackcdn.com/image/fetch/$s_!x1tS!,w_3= 6,c_scale,f_png,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack.com= %2Ficon%2FLucideComments%3Fv%3D4%26height%3D36%26fill%3Dnone%26stroke%3D%25= 23808080%26strokeWidth%3D2" width=3D"18" height=3D"18" style=3D"border: non= e;vertical-align: middle;max-width: 18px" alt=3D"">
3D""
3D""
READ IN APP=3D""
 
3D""

For a long time, execution was a very convenient excuse.

Not a fake excuse. A real one. Making things was hard, man. Building = a company was harder. Publishing consistently was hardest. Designing a prod= uct, launching a campaign, writing the essay, editing the video, building t= he landing page, analyzing the data, making the deck, sending the email, fo= llowing up, doing the second version, doing the third version after the sec= ond version revealed you were maybe not the visionary you briefly became in= the shower.

Hard.

Annoying.

Time-consuming in the specific way that makes = you look up at 7:48 p.m. and realize you have eaten one protein bar and dev= eloped a personal relationship with a spreadsheet.

So we al= l got to say some version of the same thing…

I wo= uld, but I do not have the time.

I would, but I do= not have the team.

I would, but I do not have the= budget.

I would, but I do not have the design skills,= the writing skills, the research help, the analytics setup, the editor, th= e operator, the intern, the agency, the growth person, the mysterious adult= in the room who apparently knows how to make a go-to-market motion stop lo= oking like a Google Doc fell down the stairs.

Some of that = was true.

A lot of it still is.

But AI is s= tarting to make those excuses less comfortable. Not gone. Please do not bec= ome the person who says “you can build anything now” after prom= pting a chatbot to make a todo app that stores everyone’s passwords i= n a text file named `final_final_secure.txt`. The world is still full of re= al constraints. Distribution is hard. Quality is hard. Trust is hard. Money= remains, inconveniently, money.

Still, something has shift= ed.

The gap between idea and artifact is shrinking. You can= get a draft faster. A prototype faster. A visual direction faster. A resea= rch summary faster. A first campaign faster. A rough analysis faster. A wei= rd little workflow that does the annoying first pass faster. The blank page= is less blank. The first version is less expensive. The distance between &= #8220;I wonder if” and “here is something I can react to”= is collapsing.

Which sounds like pure good news.

It is good news.

It is also going to make ambition e= xtremely weird.

when execution gets cheaper, ta= ste gets louder

The first obvious shift is that taste beco= mes more valuable. I have been beating this drum enough that the drum has p= robably filed a formal complaint, but it remains true all the same… w= hen everyone can make more stuff, the advantage moves to knowing what shoul= d exist.

But there is another layer underneath taste.

Taste is not just the ability to say “this is good” = or “this is bad.” Taste is a relationship with desire. It is kn= owing what you are trying to make real. It is having enough internal signal= to recognize when the machine produced something technically correct and s= piritually unemployed. It is being able to look at ten decent options and f= eel the one that has a pulse.

That sounds nice until you re= alize how much of modern work has trained us to avoid that feeling.

A lot of work is built around external validation. What is the ben= chmark? What is the competitor doing? What does the algorithm reward? What = did the customer ask for? What did the board like? What did the previous ca= mpaign prove? What does the dashboard say? What would be safest to ship bef= ore the meeting where everyone pretends the problem is prioritization and n= ot fear holding a reusable water bottle it received at a SaaS offsite?

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">These are not bad questions. They are necessary questions. But = they are not enough when execution gets cheap.

Because if y= ou can make ten versions, the question becomes which version do you actuall= y believe in?

If you can launch three tests, what are you t= rying to learn?

If you can generate fifty angles, which ang= le deserves your name on it?

If you can build the prototype= over a weekend, what is the product trying to make possible?

AI does not remove the need for ambition. It removes some of the camou= flage around the lack of it.

Rude technology, honestly.

=

the comfort of being blocked

Ther= e is a weird comfort in being blocked by logistics.

If the = thing is impossible because you do not have the resources, you never have t= o find out if the idea is good. If the project requires a team you do not h= ave, you never have to discover whether you were willing to lead it. If the= essay would take two weeks and you only have two hours, you can keep the e= ssay in the pristine little museum of things you would totally write if cap= italism stopped emailing you.

I am not judging this. I have= lived here. I have paid rent in this neighborhood. I have furnished the ap= artment with elaborate notes apps and the emotional support tabs of half-fi= nished ideas.

Constraints protect us from exposure. They le= t the dream stay perfect because it never has to enter the world and become= a version one with typos, unclear positioning, and one guy in the comments= explaining that technically the Romans had content engines.

It says: okay, maybe you cannot finish it perfectly today. But you can make a fi= rst version. You can test the shape. You can draft the landing page. You ca= n mock the product. You can turn the idea into something visible enough to = be judged.

Terrible.

Useful.

Terrible because now the idea has a body, and bodies are vulnerabl= e. Useful because this is where real work starts.

The dream= does not become serious when you imagine it harder. It becomes serious whe= n it has to survive contact with reality.

AI makes contact = cheaper.

That means more people are going to have to answer= a question they have been outsourcing to friction…

<= span>Do I actually want this?

productivity is not the same as direction

This= is the part that makes me nervous about a lot of AI productivity culture.<= /p>

The pitch is usually more output. More speed. More force. M= ore capacity. More drafts, more tests, more campaigns, more clips, more ana= lysis, more automation, more personal operating systems, more little agents= running around your digital life like caffeinated interns with root access= =2E

Cool.

But more output in the wrong direct= ion is not progress. It is a Peloton for your anxiety. You are moving. The = numbers are changing. There is sweat. A screen is congratulating you. You h= ave not gone anywhere.

I love systems. I love workflo= ws. I love the idea that a person with good taste and a small stack of agen= ts can do work that used to require a whole department and a conference roo= m full of snacks nobody admits to liking. This is obviously part of what we= are building toward at Averi… marketing systems that remember, lear= n, draft, analyze, and help teams move faster without turning everything in= to gray paste.

But speed only helps if you know wher= e you are going.

Otherwise AI becomes a very powerful way t= o avoid the more embarrassing question. Not “how do I make more?̶= 1; but “what am I trying to become by making all of this?”

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">That question is gross. It sounds like someone put a brand stra= tegist and a youth pastor in the same Patagonia store. I know. I hate it to= o.

But it is there.

For companies, it sound= s like, what category are we trying to define? What belief are we trying to= change? What would we still refuse to say even if it got clicks? What woul= d make this company matter if the market did not hand us relevance for free= ?

For people, it sounds like, what do I want to be known fo= r? What work feels alive enough to keep doing when the dopamine leaves? Wha= t am I willing to practice in public? What am I no longer allowed to preten= d I am waiting for permission to start?

AI makes the operat= ional questions easier.

It makes the existential ones loude= r.

Sorry.

the ambition gap

=

I think we are going to see a strange split.

S= ome people and companies will use AI to produce more of what they were alre= ady producing. More posts. More campaigns. More dashboards. More polite dec= ks with the same old strategy wearing a new font. More activity. More motio= n. More artifacts.

Others will use AI to raise the level of= what they attempt.

That is the ambition g= ap.

The first group asks, “How can = we do the same thing faster?” The second asks, “What can we fin= ally attempt because the first 30 percent is no longer such a tax?”

That difference matters.

If AI helps you pub= lish five mediocre posts instead of one mediocre post, congratulations, you= have built a small weather system of mediocrity. If AI helps you run the r= esearch, pressure-test the angle, create the draft, generate variants, anal= yze the response, and tighten your judgment until the work gets sharper eve= ry week, different game.

Same tools. Different ambition.

If AI helps you make ten landing pages because you have ten s= egments and no strategy, enjoy your content cul-de-sac. If AI helps you dis= cover which segment is actually resonating, what language they use, what ob= jection keeps appearing, and what promise you can credibly own, now the mac= hine is helping you think.

Same output category. Different = job.

The people who win will not be the ones who automate t= he most. They will be the ones who use automation to attempt something they= were previously too constrained, too slow, or too scattered to attempt.

That might be a company building a real content intelligence = system instead of a content calendar with seasonal depression. It might be = a founder finally turning scattered notes into a point of view. It might be= a creator moving from “I post when inspired” to a serious prac= tice. It might be a tiny team operating with the research discipline of a m= uch bigger one.

The size of the team matters less than the = size of the attempt.

3D""

AI makes seriousness visible

There is a= thing I keep noticing in myself and in other people: AI exp= oses seriousness.

Not talent. Not intelligence. Seri= ousness.

If someone is serious about writing, AI becomes ma= terial. They use it to argue with themselves, find holes, test structures, = get unstuck, produce clay. Then they edit. They choose. They cut. They make= the thing more theirs.

If someone is not serious, AI becom= es a costume. They ask for the finished thing, skim it, maybe change two wo= rds, and ship something that sounds like it was assembled during a layover = by a committee of autocomplete ghosts.

If a team is serious= about strategy, AI becomes an amplifier. It helps them collect customer la= nguage, test messages, build briefs, review work against principles, and mo= ve faster with more memory. If a team is not serious, AI becomes a fog mach= ine for the same confusion. More words around the same unresolved decisions= =2E

This is why I do not buy the idea that AI automatically m= akes everyone creative or strategic or dangerous. It makes serious people m= ore dangerous. It makes unserious people louder.

That is no= t a small distinction.

The tool does not decide whether you= care.

It just makes the caring, or lack of caring, easier = to see.

effort is changing shape

A common mistake is thinking cheap execution means effort stops matter= ing.

No.

Effort just moves.

Less effort on the blank page. More effort on direction.

L= ess effort on first drafts. More effort on selection.

Less = effort on mechanical production. More effort on taste, distribution, feedba= ck, iteration, and the emotional horror of looking at your own work honestl= y.

This is good. I think. Mostly. Ask me again after I have= reviewed 47 AI-assisted drafts from people who think “make it more p= remium” is feedback.

The old effort was often physica= l or logistical. Can I make the thing? Can I find the time? Can I learn the= tool? Can I get someone to help? The new effort is more psychological and = editorial. Can I decide what matters? Can I reject the easy version? Can I = keep going after the first pass looks good enough? Can I tolerate the gap b= etween the artifact and the thing I meant?

That last one is= brutal.

AI shrinks the gap between intention and artifact,= but it does not close the gap between artifact and taste. In some ways, it= makes that gap more annoying because now you can see so many near-misses s= o quickly. The machine keeps handing you versions that are almost there. Al= most there is seductive. Almost there whispers, “Ship me, coward.R= 21;

Do not trust almost there.

Almost there= is where good work goes to get a LinkedIn carousel and a quiet life.

The new effort is refusing almost there when almost there would = be enough.

permission is also getting cheaper

There is another weird thing happening.

AI = does not only reduce execution cost. It reduces permission cost.

You used to need someone to validate that you were allowed to try cer= tain things. A designer. A developer. An editor. A strategist. A person wit= h a title that made the thing feel official. This was sometimes good becaus= e expertise matters. It was sometimes bad because the gate was mostly vibes= and calendar availability.

Now you can make a version befo= re anyone gives you permission. Not the final version. Not always a good ve= rsion. But a version. Enough to show, test, learn, embarrass yourself in a = contained environment, and improve.

That changes the psycho= logy of starting.

The sentence “I do not know howR= 21; used to end the conversation. Now it starts a workflow.

I do not know how to build this prototype. Okay, ask for a scaffold.

I do not know how to structure this essay. Okay, make three outl= ines and fight with them.

I do not know how to analyze this= data. Okay, get the first pass, then verify it like an adult with trust is= sues.

I do not know how to make this campaign feel less gen= eric. Okay, generate options, compare them against your enemy list, and fin= d where the real tension is hiding.

This is not a replaceme= nt for expertise. It is a way to begin before expertise fully arrives.

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">And beginning before you feel ready is, unfortunately, where a = lot of interesting work lives.

the downside: ev= eryone can perform ambition now

Of course there is a dark = version.

If execution gets cheap, performance gets cheap to= o. People can perform seriousness at scale. They can create the artifacts o= f ambition without the substance: the pitch deck, the manifesto, the conten= t engine, the personal brand, the founder thread, the AI-generated product = mockup, the strategy memo that looks expensive and says almost nothing.

=

This is already everywhere.

The internet is fi= lling with people who have the output surface area of a media company and t= he inner life of a vending machine.

That sounds mean.

It is mean.

It is also sometimes true.

The artifacts of ambition are no longer proof of ambition. A deck is= not proof. A landing page is not proof. A content calendar is not proof. A= polished essay is not proof. AI can help make all of those before lunch, e= specially if lunch is one of those $19 salads that tastes like someone desc= ribed nutrition to a hedge fund.

So what is proof?

Consistency is proof.

Iteration is proof.

Taste under pressure is proof.

The willingness to be= specific is proof.

The ability to learn from the market wi= thout becoming whatever the market most recently rewarded is proof.

The willingness to kill your own fine work because it is not the w= ork is proof.

The artifact matters. But the pattern matters= more.

AI makes artifacts cheaper.

Patterns= are still expensive.

what to do with this

=

If you are building anything right now, I think the move is no= t just to ask, “How can AI make this faster?”

A= sk a better set of questions.

What would I attempt if the f= irst draft were free? What would I test if the prototype took a weekend ins= tead of a quarter? What would I publish if the research assistant, editor, = and repurposing machine were already in the room? What would I build if I t= rusted myself to judge the outputs instead of outsourcing the taste to the = tool?

Then ask the scarier version:

What am= I still not doing, even now that it is easier?

That is the= question with teeth.

Because if the blocker was truly logi= stics, AI helps. If the blocker was fear, identity, taste, or lack of desir= e, AI will politely generate a workaround while the real problem sits in th= e corner wearing noise-canceling headphones.

This does not = mean you need to become some maximalist productivity goblin. Please do not = turn your life into a dashboard where every sunrise is a KPI and every frie= ndship has a retention curve. The point is not to produce yourself into dus= t.

The point is to notice where the old constraints are loo= sening, and to raise the ambition instead of only raising the volume.

Make fewer things if you need to.

Make stranger = things.

Make sharper things.

Make things th= at would have been too annoying to start before, too expensive to explore, = too technically irritating to test, too messy to explain in the meeting whe= re everyone keeps saying “net new” like a prayer.

Use the machine to buy back the courage to try.

Not ju= st the capacity to ship.

the question I cannot = stop thinking about

A lot of people are going to use AI to= become more efficient versions of who they already were.

T= hat is fine. Useful, even. Many workflows deserve to be faster. Many tasks = deserve to be automated into the sun. Nobody needs to spiritually commune w= ith a CSV export at 11:34 p.m. unless that is your thing, in which case I s= upport your journey from a respectful distance.

But the mor= e interesting question is whether AI can help people and teams become more = ambitious versions of themselves.

Not louder.

Not busier.

Not more omnipresent on every channel unti= l the audience starts developing the digital equivalent of hives.

More ambitious.

More willing to attempt the thing th= ey actually mean. More willing to build the system, write the argument, def= ine the category, challenge the default, make the weird prototype, publish = the sharper take, design the better ritual, refuse the safe version, and ke= ep going after the machine has done the easy part.

That is = the part I am watching.

Because if execution gets cheap and= ambition stays small, we get more stuff.

If execution gets= cheap and ambition expands, we might get work that feels impossible from t= he old world.

Maybe that is naive.

Probably= a little.

But I would rather be naive in that direction th= an spend the next decade optimizing the production of things nobody wanted = badly enough to make well.

The machine can lower the floor.=

It cannot raise your ceiling for you.

-zc<= /p>

P.S. If this made you realize your AI workflow is very good= at helping you avoid the actual question, welcome. Horrible club. Great sn= acks.

Upgrade to paid

Subscribe for the rabbit hole for the overcaffeinated, where we keep= asking whether the machine made us more powerful or just gave our procrast= ination a nicer UI.

Rabbit hol= e for the overcaffeinated

- Ethan Mollick on holding= back the strange AI tide: https://www.oneusefulthing.org/p/on-holding-back-the-stra= nge-ai-tide

- Ethan Mollick on centaurs, cyborgs,= and the jagged frontier of AI work: https://www.oneusefulthing.org/p/centaurs-and-c= yborgs-on-the-jagged

- Andrej Karpathy’s “S= oftware Is Changing (Again)” talk, useful for thinking about what bec= omes easier when software becomes more malleable:

<= p style=3D"margin: 0 0 20px 0;color: rgb(54,55,55);line-height: 26px;font-s= ize: 16px;">- Paul Graham’s “How to Do Great Work,”= still a useful gut check on ambition, taste, obsession, and seriousness: <= /span>https://pa= ulgraham.com/greatwork.html

- Averi, where this s= hows up as the difference between making more marketing stuff and building = a system that helps teams attempt better work: https://www.averi.ai/

You're= currently a free subscriber to Don't Feed The Algorithm. For the full experience, upgrade your subscription.

Upgrade to paid

 
3D""Like
3D""Comment
=
3D""Restack
 = ;

© 2026 Zach Chmael
548 Ma= rket Street PMB 72296, San Francisco, CA 94104
Unsubscribe

3D"Get3D"S=

3D""3D"" --25e5463c370c0f1c85cf24a5210d3e83d9aaba1dd37ca78e27de70df071a--