Sam Woods

The Asymmetric Operator ==================== The operators building agents right now are mostly stuck solving the wrong problem. I ran a live briefing on this the other week for a small room of operators already building with agents. Full video attached. Short version below. Almost every agent platform is within a rounding error of the others on raw capability. The agents can do the work. Most operators have figured out the real bottleneck is coordination, so they're switching orchestration tools every other month chasing a fix. Twelve-month read: coordination is closing at the platform layer. The platforms you're using have been quietly collecting your fix-it decisions all year, and that data is the training signal for agents that self-correct. If your edge is managing agent failures, you have nine months before the default tools absorb it. The layer that compounds past that is underneath the agents. A small set of files your whole system reads from, structured as a single source of truth instead of fifteen agents each guessing from a fresh prompt. Inside that layer, the piece almost nobody is building is thought patterns, your reasoning about the work logged as you make decisions. Ten minutes a day per cohort, voice-to-text is fine. At the end of the week pull out the five or six patterns that kept showing up and drop them into the cohort's master context file. Repeat for a month and the agents' outputs start reading like a sharper version of your own judgment rather than a generic model's best guess. Six months in, the operators I advise have agents making sharper calls than any competitor's agents, because those agents have been trained on half a year of their captured reasoning. No competitor can buy that. Living through the same six months is the only way to produce it. If you want more like this every week, see the next post or profile. **Chapters:** 00:00 Why this isn't about prompts or agents 04:25 The agent landscape right now 07:45 Why agent capability is commoditized 08:30 The e-commerce friend running 80% of his business on one agent platform 10:30 What the one-person billion-dollar company actually proves 14:25 The coordination problem and why it gets solved this year 19:00 How agent platforms are quietly learning from your fixes 22:30 Scattered vs. coordinated vs. redesigned businesses 24:45 What cognitive architecture really means 28:00 Thinking in cognitive labor instead of org charts 31:10 The file layer: claude.md, agents.md, skills, memory, context, heartbeat 38:30 What makes an architecture cognitive: thought patterns 40:10 The 10-minute daily decision journal that feeds your agents 42:30 Why your thought patterns don't need to be perfect 44:40 Where to start if you don't have agents yet 46:20 Thought patterns per agent team, not per whole business 48:20 Case study: SaaS consultancy with 7 agents and 3x capacity 51:15 Decomposing work into 7 units of cognitive labor 53:30 Month-by-month build timeline 56:00 The data set no competitor can copy 1:01:20 The maturity ladder: reactive, automated, attentive, autonomous 1:03:35 What's coming over the next 12 months 1:06:40 Your job when agents handle execution 1:08:00 Why now is the cheapest this will ever be Want more? See next post

Most revenue functions I look at right now are busy without compounding. Copy ships, sequences run, deals close or don't, and the next cycle starts from roughly the same place as the last one. The team is putting in the hours. The output is constant. The brief files everyone writes from never get updated, because nobody owns updating them. The fix is structural. Every revenue function has three layers: what you read going in, what you make from it, and what you learn going back. Most operators only run the middle one. Run all three on a weekly cadence and the system gets smarter every Friday. Run one of three and it stays where it was last quarter.

An ecommerce founder I work with opened her Notion page to me last week. Title: "Marketing Ideas." It was 87 entries long. Oldest from late 2023. She'd added two more that morning, before our call. Her words: "I have all the ideas. I just don't know which one to pick up." That sentence describes the actual condition of marketing at her scale. The bottleneck is the absence of a system that converts ideas into shipped experiments. Marketing teams stall on ideas they already have. The triage is the actual work.

Most prompt libraries die in three weeks. Friction kills them. Long prompts get skipped. Specific prompts assume situations you don't have. Categorized libraries cover scenarios you'll never run. The library you'll actually use is shorter than you think.

@lingdiscovery @grok explain the functionalist approach to langauge in simple terms, why it matters, and what the implications are downstream across culture, science, writing, thinking

This is my favorite book on the evolution of language. What I love about it is that, while it does discuss both nativist (generativist) and functionalist approaches to language, I think it makes it pretty clear that only the functionalist approach really comports with all the genetic and paleontological evidence we have about the evolution of language.

as to Protestants and even Eastern Orthodox though I doubt they're interested

Nova ran a six-week audit on our team's prompt library. We had built fifty prompts across nine categories. Cross-referenced. Properly set up as saved Claude projects. Three got opened more than ten times. Five got opened occasionally. Thirty-seven were never opened at all. The survivors had three things in common: short (under 80 words), portable (worked across client types), took inputs instead of assuming the situation. The other 47 either anticipated situations we didn't have or forked on distinctions that didn't matter at the prompt level. We rebuilt the library at eight prompts. It isn't done shrinking.

A friend of mine sells a $4K productivity offer. Strong product, decent funnel, nine months of flat conversion. He had a swipe folder on his hard drive going back over a year. Six competitors. Sales pages, emails, ad library exports. He'd never analyzed any of it. We loaded the folder into a Claude project and ran one analysis pass. Twenty minutes. The pattern that came out was a single sentence. Every competitor at his price point was positioning the offer the same way: "for operators who've already tried the cheap version." His sales page hero said "for anyone serious about getting this right." The buyer arriving at the page was already pre-sold on having spent money trying to solve the problem some other way. The page was treating them like a first-attempt buyer when they were a second-attempt buyer by definition. That's the mismatch a $4K offer can't survive. He made one repositioning pass on the hero. Same offer, same price, same audience. Just the framing of who it's for. Conversion came up in two weeks. The competitor swipes had been sitting unread for over a year. The cost of reading them used to be a marketer's full week. Now it's twenty minutes. Most operators have a swipe folder. Few read it as a dataset. The leverage is in the work nobody's doing.

The tightening pass is the dangerous one. It will cut your voice if you let it. Maybe sixty percent of the suggestions are good. The rest remove the parts that make the copy yours. The slight overlong sentence that creates rhythm. The modifier that's doing emotional work. The sentence fragment that breaks the cadence on purpose. Read this pass slowly. Reject more than you accept.

@nabeelazeez money skills

claude skills this hermes skills that how about you learn some life skills

Online buyers have three conversations before they buy. They talk to themselves first, often for weeks. Then they read every competitor in the category. Last, they show up at your page. Most operators write copy for that third conversation. The first two are where the deal got decided. If you don't know what your buyer was telling themselves before they got to you, you're solving a problem that already got solved by somebody else's positioning.

@nivi tired = ads wired = podcast films

We’re making podcast films.

@thebasecreates when she's walking and talking—and specifically pointing with her hand/fingers and directs her face off camera, who is she pointing to? in the play, who is there? who does she imagine/"see" in front of her, if anyone?

Is poetry truer than prose? Project 39 continues with Mistress Quickley from The Merry Wives of Windsor - Act 2, Scene 2. The old scoundrel Falstaff is trying to seduce two ladies to get their money. But it turns out they are playing a trick on him. Mistress Quickley acts as a duplicitous go-between. It is one of Shakespeare’s lesser known plays, supposedly written at the behest of Elizabeth I as she wanted to see Falstaff in love. This speech is entirely in prose. Most of Shakespeare’s work is written in iambic pentameter verse. It is normally assumed that Shakespeare uses prose to show that a character is of a lower class. However, Giles Block - who was Master of Voice at the Globe Theatre under Mark Rylance, argues that Shakespeare uses prose when the characters are lying or trying to hide their feelings. This seems strange to us; we assume poetry to be less truthful, more fanciful than prose. But for Shakespeare and other Elizabethan writers, poetry is closer to truth than ordinary language. The duplicitous Mistress Quickley and Falstaff speaking entirely in prose here would seem to support this. Mistress Quickley is played by Naomi Preston-Low.

@TaylorPearsonMe @heyitsnoah interesting, how long have you been working on agents and AI ?

Really enjoyed chatting with @heyitsnoah on Forward Deployed about how AI is reshaping organizational structure and what knowledge work actually looks like day to day in a post-agent World. Noah pushed my thinking in a bunch of directions I hadn't gone on my own. We got into: - Why agentic systems look more like companies than software. Agents execute well but can aim at the wrong problem, which is the same issue companies have spent centuries trying to figure out. - Where the bottleneck moves once agents can write (near perfect) code. - And more! Pace layers, OODA loops, transaction cost economics, Cynefin….

@nabeelazeez amateur hour if you just think asking chadGBB for advice is legit

I had this same problem with a CMO sending me ChatGPT advice on how to improve their email marketing. It was flat out wrong. And it made me question their competence as a CMO.

@micsolana Sad!

I have started to notice very smart people who use AI every day, typically as a kind of thought partner, have started to sound like AI, not only employing whatever popular new turn of phrase, but in this kind of bulleted cadence of speaking — a copy of a copy. concerning!

@millerman what courses did you do? on what?

The most memorable, best produced, and truly impressive educational time I've had (I exclude the singularly transformative experience of reading Leo Strauss) is the courses I did at Landmark Worldwide, formerly EST, especially around "integrity". Any time I attend a conference or event that doesn't start punctually, I remember how uniquely effective those courses were (and presumably still are) in surfacing how we relate to being on time.

@davogones ah ok

@samwoods Umbrella term for autism, ADHD, etc

I want more higher quality connections If you're: - into philosophy of mind/tech - doing legit AI and Agent work (no hype psychos please) - AI agent builders - Builders/founders - love writing essays (with or without use of LLMs) - High agency person - tryna run your 7-9 figure business as best you can If any of these sound like you, say hi below

The five editing lenses I run on every draft, in order. - Clarity. Sentences that ask the reader to do too much work. - Persuasion. Sections that explain when they should sell. - Voice match. Places where the draft drifts off the target voice. - Objection handling. What the draft hasn't addressed. - Final tightening. Cut what isn't earning its keep. One pass at a time. Fresh chat for each one. About twenty minutes on a 1,500-word draft. The order matters. Each lens depends on the one before it. Full system with the prompts → copywriting.ai