How to Own a Position When Thinking Costs Nothing\u00a0<\/strong><\/p>\n<\/h2>\n
Read Time: 7\u00a0Minutes<\/em><\/p>\n Hello<\/p>\n A quick announcement before we dive into this week’s Digest. For all\u00a0Kit<\/a> users, I’ve added two capabilities to Pablo (previously GNYS “Genius”) to audit or create offers and a funnel for any brand. All you have to do is drop in the URL and let Pablo do the rest. See the image below and click on either one for a PDF output.<\/p>\n fyi. Pablo is a positioning coach.<\/p>\n Four scenarios where the audit is useful and helpful:<\/strong> Disclaimer:<\/strong> Only the Mental Monopoly System tier gets you access to Pablo (a custom GPT). You only pay once, and the access is yours forever.<\/em><\/p>\n In 2020, on one of those long lockdown walks, I told my friend Preet<\/a> that AI was going to end up like electricity. Boring. Everywhere. Something you pay a bill for and never think about. I also told him most people would one day build software by talking, and would never read a line of code.\u00a0I said nobody would be talking about AI in 2 years<\/a>.<\/p>\n The 2 years are up. People are talking about it more than ever.\u00a0I was early on the date. I still think the call lands by 2030, and I will show you what I am counting. First, I want to give you the part of this that will not move, because that is the part you can build on.<\/p>\n The internet was different. Dotcom was different. Cloud computing was different. A phone in every pocket was different.<\/p>\n All of that was true. Every one of those made a real mess on the way through, and every one of them ended up as pipework nobody mentions. Right now AI is on the same road and moving faster, and the speed is why so many sharp people cannot think straight.<\/p>\n So when the ground will not sit still, build on the parts of it that have never moved. I count 3.<\/p>\n Time.<\/strong>\u00a0Nobody has ever made more of it. Every tool since the plough got sold as a way to save time, and not one of them has added an hour to the day. Whatever AI hands back to you gets spent on something else. The only question is what?<\/p>\n Attention.<\/strong>\u00a0People do not hand it over because you asked. I wrote in September 2025 that\u00a0the short attention span is a myth<\/a>. The same person who skips 8 seconds of your ad will sit through a 3-hour conversation about something they care about. Attention has always been picky, and it has always had to be earned.<\/p>\n Refusal.<\/strong>\u00a0Everything anybody has ever built came from ignoring almost all of it. I wrote in September 2025 that\u00a0people want what is scarce<\/a>. Say yes to everything, and you leave nothing for anyone to want. Costco has held its hot dog at $1.50\u00a0since 1985<\/a>. That is 41 years of turning down a price rise that every spreadsheet in that building has asked for.<\/p>\n I wrote in January 2025 that\u00a0technology changes and human nature does not<\/a>. I have not found a reason to take it back.<\/p>\n And here is what all this speed is doing to people. Tim Ferriss wrote<\/a> that most information is time-consuming, negative, irrelevant to your goals, and outside your influence. Tim Denning added the part that lands: the more information you take in, the more you overthink basic decisions.<\/p>\n That is what I watch founders do all day now. More reading, more demos, more comparing of models, more threads about what changed this week, and fewer decisions that cost anything. Watching has started to feel like working.<\/p>\n So slow down. Every price in this piece will be wrong in a year. Every model name will be wrong sooner. Somebody still has to give you their time. Somebody still has to want to give you their attention. You still have to turn down nearly everything to be worth either one.<\/p>\n 2 words first. A model is one of these AI systems, like ChatGPT. A token is a chunk of text, about three-quarters of a word. It is how these companies bill you, the same way the power company bills you by the unit.<\/p>\n GPT-3 launched in November 2021 at $60 for a million tokens. 3 years later,\u00a0the venture firm a16z<\/a>\u00a0found a small free model scoring the same on the same test for 6 cents. That is a thousandfold drop in 3 years.\u00a0Epoch AI<\/a>, which tracks this for a living, puts the fall at about 40 times a year for a given level of thinking.<\/p>\n The people selling it say so out loud.\u00a0Sam Altman, last July<\/a>:\u00a0“It looks like we are about to deliver on intelligence too cheap to meter.”\u00a0Satya Nadella, in December<\/a>: “In some sense, models are becoming a commodity. A commodity is something you buy on price because every version of it is the same.”<\/p>\n They are also getting hard to tell apart. There is a public leaderboard where people vote on which answer is better.\u00a0Stanford’s AI Index<\/a>\u00a0looked at it this year and found the top 4 models sitting almost on top of each other. The best American model is 2.7% ahead of the best Chinese one. DeepSeek sells its cheap version for\u00a014 cents a million tokens<\/a>. Marc Andreessen called the free models anyone can download an\u00a0asteroid strike<\/a>\u00a0on the money the big labs were counting on.<\/p>\n The buyers already moved. When Andreessen Horowitz\u00a0asked the tech bosses of large companies in January<\/a>, 81% said they use 3 or more AI suppliers, up from 68% a year before.\u00a0Perplexity published its own numbers<\/a>: 2 models did 91.5% of the work in January 2025, and by December no single model was doing more than 23.3%.<\/p>\n Look at what happened at the top of\u00a0Anthropic’s own price list<\/a>. Its best model cost $15 in and $75 out per million tokens in mid-2025. In November 2025,\u00a0Opus 4.5 arrived at $5 and $25<\/a>. That is a 3x cut at the frontier. Then 4.6, 4.7, 4.8 and Opus 5 all came out at that same $5 and $25, each one better than the one before it. The price page today also carries 2 newer models above those, Fable 5 and Mythos 5, at $10 and $50, and even that premium tier costs a third less than the best model cost a year ago.<\/p>\n So the price of the best thinking money can buy fell by two-thirds and then sat still while the thinking got better 5 times over. That is a utility behaving exactly like a utility.\u00a0Nobody is going to win this by picking the smartest model. Build so you can switch in an afternoon, then put the saved time into what nobody can rent from you.<\/p>\n Right now, “AI” is everywhere. 4 times a year, big companies hold a call to tell investors how they did.\u00a0FactSet counted the word on 337 of those calls<\/a>\u00a0at America’s biggest 500 companies in early 2026. That is the most in the 10 years they have counted, against a normal year of 103. Companies that said it watched their share price rise 12.7%. The quiet ones got 2.6%. Meta named a whole division Superintelligence Labs.\u00a0TechCrunch reported ChatGPT going<\/a>\u00a0from 400 million people a week to 900 million.<\/p>\n Now look at what those same companies sell. Apple got 28 minutes into its June event before saying “AI” once,\u00a0which CNET counted<\/a>.\u00a0The Wall Street Journal reported<\/a>\u00a0Apple, Google, Microsoft and Samsung all pulling or rewriting AI claims in their ads after the advertising watchdog took a look.\u00a0The Guardian reported<\/a>\u00a0Apple paying $250 million over Siri features it advertised and never delivered.\u00a0Reuters reported<\/a>\u00a0Starbucks switching off its AI stock counter in 11,000 shops after 9 months because it kept mixing up oat milk and dairy. Merriam-Webster made “slop”\u00a0its word of 2025<\/a>.<\/p>\n Think about your last power bill. You paid it without reading it. You have never once thought about electricity while filling a kettle, and no kettle has been advertised for its electricity in your lifetime. Meanwhile, the power company is on the news every time the grid goes down, and in front of a regulator every year asking for more. The selling stopped decades ago. The spending never did.<\/p>\n That is the shape I should have described in 2024. I said the word would go quiet. The word is going quiet on products first, and I think it is mostly gone from how things get sold by 2030.<\/p>\n Here is a count you can run this week. Take the 10 biggest companies in your world. Count how many say “AI” on the pages where they sell things. Then count how many say it in the reports they file for investors. Do the same on yourself. If you say it more to customers than to your accountant, you are selling the machinery instead of the result.<\/p>\n Yahoo Finance added up<\/a>\u00a0what Amazon, Microsoft, Alphabet and Meta say they will spend this year on buildings, chips and power for AI. It comes to about $725 billion. Last year the 4 of them spent $410 billion. That is nearly double in one year.<\/p>\n Some of that bill is being made to look smaller. On July 29,\u00a0SiliconAngle reported<\/a>\u00a0Microsoft’s finance boss telling investors the company now assumes its buildings last 25 years instead of 15. Stretch the life of a building and this year’s share of the cost shrinks on paper. She said in the same breath that the cash leaving the company does not change. Michael Burry, who is betting against these companies and is worth reading anyway,\u00a0told CNBC<\/a>\u00a0that the move is one of the more prevalent frauds in contemporary finance.<\/p>\n The payoff has not landed.\u00a0MIT looked at companies trying AI last August<\/a>\u00a0and found about 95% of those projects made no difference to profit.\u00a0Chamath Palihapitiya, talking about the electricity side in May<\/a>, said less than half of it is actually being built, and most of it is stuck in red tape.<\/p>\n Jeff Bezos said the useful thing in Turin last October<\/a>. AI is a horizontal enabling layer, meaning everything ends up running on top of it, and what is happening now is a kind of industrial bubble. Both are true at once. Railways ended up boring and useful, and they ruined most of the people who laid the first tracks. Webvan had home grocery delivery right in 1999 and was dead by 2001. Instacart did the same thing 10 years later using shops that already existed and drivers’ own cars, and it worked. Edison built the generator and lost control of his company. GE got rich making the things you plug into the wall.<\/p>\n Knowing where this ends up is worth nothing on its own. The road is what kills you.<\/p>\n Every new technology spends its early years shaped like the people who built it. Early cars came with a crank and a toolkit, so you had to be part mechanic. Early electricity got sold as wiring. Early computers made you type commands in a language you had to learn first.<\/p>\n AI showed up as a text box. Then a whole industry appeared to teach people how to talk to the text box. That is the giveaway. “Learn to prompt” is the same instruction as “learn to code.” Both mean the same thing. Think like the person who built this, because the person who built this is not finished.<\/p>\n Andrej Karpathy saw it in January 2023 and wrote one line, that\u00a0the hottest new programming language is English<\/a>.\u00a0In February 2025 he wrote<\/a>\u00a0about how he now forgets the code is even there. Sundar Pichai told investors last year that\u00a0more than 30% of new code<\/a>\u00a0at Google is written by AI and checked by an engineer. Nadella says 20% to 30% at Microsoft. Everyone counts this differently, so take those as a direction.<\/p>\n 3 weeks ago the next step arrived. OpenAI put out\u00a0a new voice model<\/a>\u00a0on July 8. On July 23 it went into the desktop app, and\u00a0Fortune reported<\/a>\u00a0you can now build software by talking to it while it works in the background.\u00a0TechCrunch reported Anthropic releasing its own version<\/a>\u00a0the same week across Gmail, Calendar, Slack and Notion.<\/p>\n Cars lost the crank. Phones lost the operator. Computers lost the typed commands for everyone except the people who needed them. Code is next. By 2030, most people will build working software without reading a line of it, and code becomes a specialist’s material like sheet metal or wiring. The engineers who read it will get paid more, because anyone who understands what sits underneath gets rare the moment everyone else stops looking.<\/p>\n Open the first 10 minutes of using your product and delete every instruction that teaches a customer how to talk to it. What is left is your actual product.<\/p>\n Researchers at the Wharton business school ran 5 brainstorming experiments and\u00a0published them in Nature Human Behaviour<\/a>. People working with ChatGPT had better single ideas and far fewer different ones as a group. In one task, 94% of the ideas overlapped, and 9 people separately gave their invention the same name. Rory Sutherland got there first in\u00a0his book Alchemy<\/a>, where he wrote that logic always gets you to exactly the same place as your competitors.<\/p>\n Which brings up the oldest line in business, that A students end up working for C students. The evidence is better than the folklore. In 1921 Lewis Terman tested 168,000 California schoolchildren and followed the 1,528 smartest of them for life. His colleague Melita Oden compared the 100 most successful men in that group with the 100 least successful.\u00a0Stanford’s own magazine reports<\/a>\u00a0their test scores came out almost the same. Confidence and stubbornness did the separating.<\/p>\n Torkel Strenze pulled together decades of studies<\/a>\u00a0and found the link between test scores and income is real and small.\u00a0Jay Zagorsky ran a big American survey in 2007<\/a>\u00a0and found no measurable link at all between test scores and how much wealth somebody ends up with.<\/p>\n That is my read, and to be fair, there is a second way to read it. Sweden used to test every young man who signed up for national service.\u00a0A 2023 study of those records<\/a>\u00a0found the high scorers earning about 20% more in wages. It also found them earning about 60% more from owning things like businesses, shares and property. That held after the researchers allowed for schooling, job, savings, inheritance and parents. So the numbers alone cannot settle whether brains buy you the outcome or just buy you the odds. I lean on the Terman result, because it holds the brains steady and watches everything else do the separating. Anyone telling you intelligence does not matter is either hiding or selling you something.<\/p>\n So brains are not the problem. Brains buy you choices, and choices are not a direction. The C student picked one before anybody gave him permission, then hit the hard problems one at a time, in the order the work handed them over. The A student can see every one of those problems on day one and prices the odds correctly, and seeing the odds clearly is what stops him. Then the C student hires the A students and points them somewhere. That was always the deal. The only new part is that the A students now also rent for 14 cents a million words, with no notice period.<\/p>\n Write down every choice you are keeping open. Cross out the ones you are keeping open because you cannot decide. What is left is your real to-do list.<\/p>\n Here’s what I actually use. A position moves through 8 states, in this order.<\/p>\n 1. Conviction.<\/strong>\u00a0A founder’s standard (a belief about a future reality)<\/em>, held before there is a customer or a set of accounts to test it against. Free, easy to walk back, invisible from outside.<\/p>\n 2. Concept.<\/strong>\u00a0One noun, chosen. Adjectives do not lodge in memory. Two nouns is zero nouns.<\/p>\n 3. Sacrifice.<\/strong>\u00a0The first move that costs something. Money refused, a type of customer turned away, a working product killed. Everything before this is a preference. This is where a preference turns into evidence.<\/p>\n 4. Structure.<\/strong>\u00a0The business (operating model) <\/em>bends around the noun. Where the money goes, what it costs to run, who talks to whom, what gets built. If the noun vanished tomorrow and nothing in the company would have to change, the noun was decoration.<\/p>\n 5. Proof.<\/strong>\u00a0Verbs, not adjectives (what people can verify)<\/em>. Receipts a stranger can check without your permission. Other people’s voices, which carry more weight than anything you will ever say about yourself.<\/p>\n 6. Repetition.<\/strong>\u00a0Around 80% of the places a customer meets you saying the same thing, for years. Not a campaign. A decade of not contradicting yourself.<\/p>\n Those 6 are yours. You can pay for them, put them in the calendar, check them, and fire somebody over them.<\/p>\n 7. Association<\/strong>\u00a0is not yours. It happens inside a customer’s head, on their timeline, with no input from you. There is no way to install it.<\/p>\n 8. Position<\/strong>\u00a0is not yours either. It gets granted. The market decides your company and the noun are the same thing, and it decides late, quietly, and without telling you.<\/p>\n Now put cheap AI against that list.<\/strong><\/p>\n It made states 2, 5 and 6 nearly free. Anybody can produce a noun, a page of proof-shaped copy and a year of posts before lunch. That is what the flood in your feed is.<\/p>\n It did nothing at all to state 3. Sacrifice is the one state whose whole value is what it costs you. Turn a customer away and you lose that money whether or not AI wrote the announcement. There is no cheap version of a decision that hurts.<\/p>\n It cannot reach 7 or 8, because those happen inside other people’s heads.<\/p>\n Everything AI can make for you is now the price of entry. What you paid to not do is the only thing a competitor cannot match by Friday.<\/p>\n This is also why changing your words never works. Companies redo state 2, skip 3 through 6, and wonder why 7 and 8 never arrive. Words come last. They describe what the business already decided. Put them first and you are asking people to believe a sentence nobody paid for.<\/p>\n You do not create a position. You pay for one, and the market decides whether it was enough.<\/p>\n Write your noun. One word. If you need 2, you have none.<\/p>\n List every decision from the last 12 months that cost you money and only makes sense if that noun is true. Money turned away. A customer let go. A feature killed. Somebody you refused to hire. Put the amount and the date beside each one.<\/p>\n Then list every decision from the same 12 months that argues against the noun. Money taken. A customer served who does not fit. A product released that any competitor would also release.<\/p>\n Read the 2 lists. An empty first list means you are sitting at state 2 with a nicely written website. Entries on both sides mean you are paying for a position and undoing it at the same time. A full first list and an empty second one is rare, and it is the whole game.<\/p>\n Last, write the 3 things you will not do for the next 12 months and what each one costs you. A refusal that costs nothing is a preference, and preferences get copied by Friday.<\/p>\n 3 exits. Tell me when you see one.<\/p>\n One company gets so far ahead on quality that price cannot catch up for 2 years or more. Then these are products and not a utility, and I was wrong about the shape of all of it. Anthropic holding at $5 and $25 since November 2025 is the first sign, and it is the one I check most often.<\/p>\n Moving gets so hard that nobody moves. Say your files, your tools, your history and your permissions all sit inside one supplier. Then this is a utility the way the giant shared company computer of 1975 was a utility, which is to say it is not one.\u00a0AgentMarketCap ran a study in July<\/a>\u00a0and found 46% of large companies naming the cost of moving as the thing blocking them.<\/p>\n The money breaks before the technology gets boring. Then this decade ends in half-built data centres and ruined companies, and the destination was right down a road that killed the people who took it.<\/p>\n The date is the loosest part of all of it. I said 2 years, I now say 2030, and that is exactly why I gave you something to count instead of something to believe.<\/p>\n Stop shopping for the best model and make yours swappable in an afternoon. Run the word count on your 10 biggest competitors and on yourself. Write your one noun. Then name the 3 things you will refuse for the next 12 months and put a price on each. Those 3 lines are the only part of this that AI cannot do for you, or for anybody else.<\/p>\n I am watching 3 things over the next year. Whether Anthropic’s top price finally moves. Whether the cost of switching climbs high enough to lock people in. Whether the money holds long enough for the boring part to arrive. The middle one is the one I would most like to be wrong about.<\/p>\n PS\u2014If you want the noun and the refusals worked out in an hour instead of a quarter, that is what\u00a0Clarity Kit<\/a>\u00a0is for.<\/p>\n Access 30+1 Positioning Reports<\/a><\/p>\n There are three ways to go deeper, depending on where you are.<\/p>\n 1. Read and follow.<\/strong>\u00a0 2. Use the systems.<\/strong>\u00a0 3. Work together.<\/strong>\u00a0 DIGEST<\/strong><\/p>\n Memo<\/em><\/a>\u00a0| Read previous digests<\/em><\/u><\/a><\/strong><\/p>\n","protected":false},"excerpt":{"rendered":" \u265b 04.08.2026 How to Own a Position When Thinking Costs Nothing\u00a0 Read Time: 7\u00a0Minutes Hello A quick announcement before we dive into this week’s Digest. For all\u00a0Kit users, I’ve added two capabilities to Pablo (previously GNYS “Genius”) to audit or create offers and a funnel for any brand. All you have to do is drop […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_coblocks_attr":"","_coblocks_dimensions":"","_coblocks_responsive_height":"","_coblocks_accordion_ie_support":"","_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"{title}\n\n{excerpt}\n\n{url}","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"_wpas_customize_per_network":false,"rank_math_title":"","rank_math_description":"","rank_math_canonical_url":"","rank_math_focus_keyword":""},"categories":[91],"tags":[],"class_list":["post-90061","post","type-post","status-publish","format-standard","hentry","category-digest-archive"],"jetpack_publicize_connections":[],"jetpack_sharing_enabled":true,"jetpack_likes_enabled":true,"jetpack-related-posts":[],"jetpack_featured_media_url":"","_links":{"self":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/90061","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/comments?post=90061"}],"version-history":[{"count":0,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/90061\/revisions"}],"wp:attachment":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/media?parent=90061"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/categories?post=90061"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/tags?post=90061"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}![\"\"](\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/10\/2d27ac01-1be3-3acf-800e-db09b618bedf.png\")
<\/a><\/p>\n
– On your business (see your blind spots and an outside-in market perspective)<\/em>
– On your existing clients (especially if you are a services business)<\/em>
– On prospective clients (perfectly suited for calibrating great sales outreach, early conversations)<\/em>
– On the market\u00a0(if you’re trying to understand the competitive landscape)<\/em><\/p>\n![\"\"](\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/10\/339836b6-99fe-57c8-03fe-cca6c49334dd.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/10\/1890a0a0-0d4e-d7b6-f258-28c2616640aa.png\")
<\/a><\/p>\n\nWhat Stays Expensive When Thinking Gets Free<\/strong>
\n<\/h2>\n\nEvery time somebody says this time is different<\/strong>
\n<\/h2>\n\nThe price of thinking fell off a cliff<\/strong>
\n<\/h2>\n\nCount where the word shows up, not how often<\/strong>
\n<\/h2>\n\nBeing right about the destination protects nobody<\/strong>
\n<\/h2>\n\nThe text box is temporary<\/strong>
\n<\/h2>\n\nEverybody asking the same thing gets the same answer<\/strong>
\n<\/h2>\n![\"\"](\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/10\/7a14fad9-cf67-8d26-186c-0b1e01c1d623.png\")
<\/p>\n\nThe 8 states<\/strong>
\n<\/h2>\n\nOne page, one sitting<\/strong>
\n<\/h2>\n\nWhat would make me wrong<\/strong>
\n<\/h2>\n\nWhat does this mean for you?<\/strong>
\n<\/h2>\n\nIf any of this was useful
\n<\/h3>\n
Everything I publish is free. The Gravity Reports, the Digest, the positioning frameworks \u2014 all of it is on the blog<\/a> and in your feed and inbox.
LinkedIn<\/a>\u00a0or\u00a0X<\/a><\/em><\/p>\n
I\u2019ve built a set of diagnostics for founders and brand leaders who want to do this work on their own company. Start with the free Analyzer \u2014 it shows you the gap between what you think you sell and what customers actually buy, and tells you whether AI recommends you when buyers are searching in their language. If you want to go further, the Clarity Kit and Monopoly are built for the same diagnostic, at a deeper level.
Start with the free Analyzer<\/a><\/em>
CEO\u00a0Clarity Starter Kit<\/em><\/a>
Monopoly<\/a><\/em><\/p>\n
I take on a small number of advisory clients each year. These are founders and leadership teams who want an outside read on their position \u2014 the same method as the Gravity Reports, applied to their own company. If that\u2019s relevant, the best place to start is a conversation.\u00a0Better call Paul.<\/em><\/a><\/p>\n