Anyone building with AI has tried the obvious fixes: a newer model, a bigger context window, a fourth rewrite of the prompt. The output got smarter and still sounded like everyone else, because everyone rents the same models. What's different is everything around the model.

The formula AI engineers use: Agent = Model + Harness.

The large language model (LLM) is the brain you rent. The harness is everything you wrap around it: the binder every new hire gets in week one, plus the editor who reads their draft before it goes out. I'd argue the harness now decides more about whether AI works for you than the model does.

Your AI already runs on a narrative harness. You just didn't write it.

The Anatomy of a Narrative Harness

Most of an AI harness is plumbing you or your engineers already build: tools, permissions, and runtime.

The Narrative Harness is the part that holds meaning: what you believe, the words you use, how you sound, and the checks that hold your AI to all three.

It runs at three altitudes:

1. At thesis, the CEO writes and ratifies strategy, worldview and big bets

2. At execution, functional heads define words, decision rules, and sign-off

3. At encoding, whoever runs AI infra turns it into one system of record every tool reads

Let’s follow one decision down this waterfall:

At thesis, the CEO rules that the company sells a platform to the mid-market. At execution, the head of product marketing writes it into the lexicon map: the product is a "platform," never a "tool," a "solution" or an "app." At encoding, that ruling lands in one versioned file every agent loads before it writes, and a linter, a check that runs on every draft, flags "tool" before it ships.

Under all three sits the substrate: one source of truth, ambient in the background, raising the floor so everyone's singing from the same song sheet. When a draft comes back saying "tool," the fix goes into the source, never into one more prompt.

What Your Context Layer Can't Tell Your AI

In October McKinsey told CEOs to build an AI operating system.

They said: "Own the context; rent or replace the technology."

They're right that every company needs a context layer: every email, Slack thread, doc and call transcript, piped in and searchable. It just isn't enough.

Watch what that does to one noun. Sales' AI calls your product a platform. Product's AI calls it a tool. The board deck calls it a solution, and the onboarding emails call it an app.

Your CEO did decide, once, in a leadership meeting in January: it's a platform. Your AI has seen "tool" 4,000 times across your Slack and your docs. Guess which one it writes.

The coherence layer holds every ruling like that, everything your company has decided it means: your worldview, your first principles, the words that mean something specific here, and how you sound. That part usually gets left to whoever happens to write the prompt that day.

I call it language debt. It's only gotten more expensive since I wrote about it.

Language debt is the new technical debt.

It compounds the same way. Every word nobody ruled on taxes alignment a little more each time it gets copied, and AI made the copying free.

That’s because AI amplifies clarity, and AI amplifies confusion. It chews through all language left unattended.

The cleanest evidence comes from Anthropic's own data team. Giving Claude raw retrieval access to thousands of prior queries moved accuracy on their analytics evals by less than a percentage point.

Writing the domain down, in plain reference files that say what each table covers, which filters apply and what the domain means in plain words, took accuracy from no better than 21% to above 95%. Their rule: let Claude write the documentation, and have "a human own the definition."

Metrics are the easy case. A revenue number has only one right answer. Your strategy and your voice have dozens of plausible ones, and a model with nothing to rule between them picks the average.

Five Words Every Narrative Harness Runs On

These five are what engineers talk about for making AI reliable.

You don't have to build any of them. You do have to know what they mean, because each one is a decision your company either made or left to chance.

AI can draft most of this. Nimit Mehta, who runs the ontology company TopQuadrant, put it plainly: "AI can take you from zero to eighty percent of an ontology faster than ever before... The last twenty percent is where it goes south, because that is where humans have to sit in a room and agree."

A Narrative Harness in the Wild

We built our own narrative infrastructure first.

Over the last 18 months, Storied's narrative harness grew into 14 git repos: Internal OS (our worldview, methods and voice rules), Content Flywheel (how we make content), our skills directory, the Storied AI MCP server and the services around it, and a few public GitHub repos anyone can install.

After being quiet in stealth about this, I’m now publishing what we’re learning.

Our Internal OS alone holds about 2,000 files, including philosophy, first principles, words we've banned and the AI tells we check for. 50+ Narrative Intelligence skills run on top of it, and about 500 commits since May show who changed which rule, when and why. Every draft that leaves Storied, this issue included, is written on that harness and checked against it after. With lots of human chiseling and editing in between.

When Ameno, our AI Platform Architect, first walked through it, his reaction was "holy #$%#$!" He'd expected brand guidelines. He found an ontology for strategy and business decision-making. This led to our Storied AI MCP.

The same narrative harness work is now part of how we partner with clients:

• A regulated science company. As AI agents start running parts of the science, the people who own compliance have become part of the harness. They’re the confidence and compliance layer for agentic science.

• An enterprise software company mid-disruption. Its category was breaking and its GTM is being rebuilt around agentic products. The harness holds one meaning for what the company is while everything around it moves.

• A philanthropic venture fund raising $100M+. Its investors range from Silicon Valley tech to conservative business leaders. The harness carries the overall conviction across multiple worldviews, and translates the pitch into the language of love for each ultra-high net worth individual.

They’ve all built the antidote:

A company that has decided what it means, and its AI won't let anyone forget it.

Who Owns Your Narrative Harness?

My call: by the end of 2027, companies running AI at scale will have a named owner of their narrative harness, and "who owns it" will be a standing line in AI governance reviews, next to data access and model risk.

Don’t worry, you can also build a narrative harness at the n=1. Just simply for your own single-player use to advance your career, ideas, and influence.

The data world was the first to realize ontology is everything.

Anthropic's rule, a human owns the definition, already reads like a job description. “Who owns meaning” comes next, and its ownership splits the way the harness does: the CEO owns the conviction, functional heads own the working definitions, and a tech operator owns the system of record.

Palantir built a public company on the concept of ontology. Its docs describe its Ontology as "a digital twin of the organization," built "to facilitate better decision-making in an organization at scale." In the second quarter of 2026, its US commercial revenue grew 149% year over year. Whoever holds the ontology holds what the company means. The question for you is whether yours lives inside a vendor's platform or in a repo you own.

Every narrative harness also needs someone who holds the whole of it: that's the role of the Narrative Architect.

Maybe bigger context windows make all of this moot. I'm betting they won't.

Ten Minutes to See the Harness You're Running On

Run this audit on Monday.

It will take 10 min. It walks three questions: what loads before your AI writes, who rules on it, and what checks it after.

See what loads. Paste this into the AI that knows your work best:

Write out the narrative harness you currently use when you write about my company, in two parts. Thesis: what we're betting on next year, what must be true for that bet to pay off, and what we believe about our market. Execution: our three first principles, the five words we use in a specific way and what each one means, how we make our most common decision, and how we sound. Mark every line with where it came from: a document I gave you, our conversations, or your own assumption about companies like ours. Be blunt about the assumptions.

Rule on it. Send the result to two people on your leadership team with one question: is this us? Then count the lines marked "assumption." That's the part of your harness somebody else wrote.

Check it. Have three people, a functional head, a recent hire and someone who sells, each ask their own AI to explain your big bet in 100 words. If the three answers don't match, fix the source, never the prompt.

Your AI already runs on a narrative harness. Write yours.

If you already have a harness or want to build one yourself, apply for access to our Storied AI MCP beta. If you want a narrative harness running across your company, that's the narrative infra work we do. Just DM me.

What do you want me to dig into next? Tell me in the comments.

Now, go own your story.

Michael

It would take 30-50 hours to build a strategic narrative.

This was for executive reviews, in front of Mark and the management team.

Every word of every sentence had a job to do.

We needed to sell the future and validate the past. Usually an abstract concept backed by user data from billions of people. You had to find a throughline, an arc, a way of driving decisions and making meaning at the boardroom level.

One group I worked with was a 1,500-person org stitched together through a reorg. Their charter was to accelerate growth through partners, the largest internet companies and marketing service providers in the world. It was a massive operation without a unified mission, much less a shared language. And it had to explain its value to the COO.

Getting the narrative right made that team’s work clear, compelling, and legitimate. It also propelled the VP to a prestigious next-level promotion.

Today, with Storied AI, I get a working narrative to 80% in 30 minutes.

I’ll take that trade any day.

The catch is the other 20%.

AI can’t write that part. It’s you: what you believe, said in your own words. By the end of this email, you’ll have a way to get yours down. It takes 15 minutes to get started.

Welcome to the Age of Narrative Intelligence. Narrative is the source code of generative AI, the prima materia. And AI is what you make it.

AI is your friend. You just have to train your dragon.

So what are you training yours on?

Sovereign AI for One Company, or One Career

Sovereign AI usually means a country building its own models and compute, so its AI runs on its own language, laws, and values.

The same logic works for one company or one person’s career.

In the Age of Narrative Intelligence, it comes down to five moves:

• Own it. Know what you believe and the bets you’re making

• Author it. Write it so a colleague would know it’s yours with your name covered

• Co-create it. Have AI argue the other side and find the gap

• Share it. Listen for your team and customers repeating it in your words

• Install it. Put it where your AI reads it first as part of your voice and thinking

All five start in the same place: a document your AI can read.

Write Down Who You Actually Are

Start where I did:

A User manual for how you think, decide, and communicate.

It’s a move I first learned from Julie Zhuo, ex-VP of Product Design at Facebook.

Your co-workers have spent years building a model of you by watching you work. Your spouse and your kids have too. Your AI has been watching you too. In fact, it probably knows you better than most in your life. You just have to ask it to reflect and mirror back. Until you ask, it keeps that picture to itself and writes like a generic composite of a leader.

That’s why you want to codify your thinking.

My CEO User Manual is a free Storied skill on GitHub, and despite the name, it works for anyone who leads a team, a project, or their own career. Mine’s the sample in the repo.

Paste the prompt into whichever AI already knows how you work. It drafts your first version on its own, from what it’s already seen: your chats, your docs, the way you write. Then it asks five questions to sharpen it. Fifteen minutes gets you started. Yes, AI writes the first draft of you. Your 20% is what you change. Then put it in your custom instructions or a Claude Project.

If you find it of value, star the repo, you’ll make my kids proud.

Before you rely on it, hand it to a colleague who knows you well. If they say “that’s exactly you,” you’re done. If not, they’ll tell you where the story feels a little off.

Your AI will only ever be as human as you’re willing to be on the page.

Michael

What it means to be a storyteller, though, is changing.

AI can draft a decent story in seconds. Deciding which story the company tells often happens in meetings most storytellers aren’t invited to.

Truth be told, I’ve never fully identified with the frame of “storyteller.” Instead, I’ve always been about the narrative: the business decisions a story has to serve.

My work lives upstream, with CEOs, CMOs, CPOs, CTOs, investors, and the companies they back, where the words that drive a decision get treated like code.

Too often, storytellers work downstream of someone else’s decision.

• The CEO sets the strategy

• The board approves the investment

• The product team locks the roadmap

• The head of marketing defines a campaign

And then it lands with the storyteller:

Get people to invest, fund, buy, or believe. That’s real craft.

Inside the company, that’s comms or marketing. Outside it, it’s the agency, the consultant, the advisor. Either way, the brief arrives after the call’s been made.

If you believe in the power of storytelling, you want to be brought in while the decision’s still taking shape.

That’s what a Narrative Architect does.

Narrative Architect: Storytellers make the stories.
Narrative Architects build and manage the language system a company runs on.

A company’s language system is its worldview, beliefs, principles, norms, values, strategy, offerings, and history, codified into a canonical stack that people and AI agents can both run on. When it’s written down, a new hire, a board member, and an AI agent all describe the company the same way.

When it isn’t, you get three versions of the strategy and a chatbot inventing a fourth.

Why Fluency Moves You Upstream

Every decision gets made in a language. Speak it, and you get a seat at the table. Don’t, and you hear about the decision after it’s made.

Here’s my contrarian take: storytelling structure is the poker ante.

It’s the price of admission. Everyone at the table expects you to frame the conversation, paint the picture, humanize something abstract, make it clearer, make people feel it, and make it stick.

Every world has its own language.

The CFO speaks in payback period and downside. The product team speaks in roadmaps and user research. Sales speaks in deal velocity and pipeline. Your AI team speaks in evals and harnesses. And when a reorg or AI transformation asks people to work differently, behavior change has its own language too.

Learn the language of the world you want to be a part of.

Fluency Lets You Call Bullshit on AI

AI can make anything sound right.

Fluency is how you can tell when it’s close, but no cigar.

Ask a model to write the case for a product launch, and you’ll get something like this: “This launch will drive strong growth and deepen customer engagement.” A CMO, much less a CFO, can’t really fund a word of it.

Someone fluent in capital and ROI writes this instead: “This launch pays back in 14 months if churn stays under 3%, and if we’re wrong, it costs us the Q3 roadmap.”

The first sounds true. The second is either true or it isn’t, and you’ll know which. That distinction only comes with fluency.

Where Each Language Gets Written

Every one of these languages lives in a document.

Starting with the ones closest to the money:

• Capital. The budget, investment memo, board deck

• Strategy. FY27 strategy, OKRs, an offsite pre-read

• Go-to-market. The funnel dashboard, pipeline review, win-loss rates

• Product. The roadmap, PRD, user research, PLG motions

• AI transformation. Architecture, evals, system prompts agents run on

• People and change. The pulse survey, change plan, adoption dashboard

Few people are fluent in all six languages.

You don’t need to be.

One more is enough to get you in on the decision earlier.

Your Next Language

Pick the one you’d like to learn next.

Open Claude or your preferred LLM and paste this:

I want to become fluent in the language of [capital, strategy, go-to-market, product, AI, or people and change]. Attached is a document I need you to translate for me into this world. Identify the 10 most important frames or concepts that will have currency for the world I want to influence, and tell me what decision this document needs to argue for. Then quiz me until I can explain that decision back in its own language.

No document handy? Ask it to write a realistic one for a company like yours first.

Then give it something you wrote this month and ask what sounds right versus what’s actually right in that language. Start with capital, and you’ll find out why the launch got funded but the rebrand didn’t.

Everyone is still a storyteller. The question is what level of the game you’re playing.

The next language you learn decides how early you get asked in.

Michael

Want to become a Narrative Architect? Storied AI puts 50+ Narrative Intelligence® skills inside Claude, and the private beta is open by request. Join the waitlist

Almost 100 applications in 24 hours, from one LinkedIn post.

Some of the best people I know read this newsletter, so before this seat fills, I want you to see it.

Storied builds narrative as infrastructure: a company’s language, managed like a code base. Our clients include VC and PE funds, enterprise SaaS, startups raising capital, growth stage firms, global life sciences, and civic ventures.

The role:

• Co-facilitate sessions with me and our executive clients

• Run the weekly cycle on narrative, messaging, and client artifacts

• Work with our narrative AI stack to improve how you think, build, ship

• Ensure every client deliverable is high-craft, executive-ready

• Coach and help manage two other members of our client team

• Help improve and spread our open tools, like Storied AI MCP

• Remote, on retainer, no business development

• Part time, 20 hours a week

The share of LinkedIn job postings asking for a “storyteller” grew 2x last year, to 70,000 listings according to the Wall Street Journal. Anthropic, OpenAI, Google, Netflix, and Microsoft are all hiring these roles. Those seats tell one company’s story. This one builds the picks and shovels for the Narrative Architect as a new category of storytelling professional.

Our role is built for a senior comms leader who’s left an in-house role to run their own practice. You’re a strong operator and C-suite partner. You just want to do the work, not have to sell it.

If that’s you, apply here.
Or share this with a person or two who come to mind.

If we hire your referral, I’ll send you $500 of the world’s rarest craft chocolate. If you know me, you know how holy f*ckballs that gift will be.

By 2029, the most sought-after operator in comms will be the one who can run narrative as infrastructure.

Michael

P.S. Not looking for a role, but want to become a Narrative Architect yourself? Hit reply and type one word: architect.

You can usually see it in the strategy.

It lives in a doc somewhere on a shared drive.

And it’s called “Strategy_v8_newCEOversion_almostfinal_vF”.

• Version 3 went to the board in the spring

• Version 5 is still in the sales team’s pitch deck

• Version 6 got uploaded to AI last month, and everyone’s using it

• Version 8 is the one that’s actually true, but it lives mostly in the CEO’s head

You also see it when someone new joins the team.

They ask four people what the company is going after this year and get four answers that don’t match. Sales gives the version from the last customer call. Product gives the roadmap. And the CEO gives whatever got decided on Tuesday.

If you asked five people on your team, how many versions would you get?

None of this is anyone’s fault.

The strategy got rewritten in March and again in August, while the documents stayed where they were. The market moved and the AI strategy evolved.

I see it with founders raising their next round and with leadership teams at global companies resetting corporate strategy for next year. A plan that used to hold for a few years now changes every few months.

That’s how things get lost in translation. The same word ends up meaning one thing to sales and another to product.

No single source of truth. The faster the plan changes, the wider the gap gets.

What Happens When You Hand It to AI

AI runs on documentation.

Traditional software is deterministic: same input, same output, every time, which is why you trust a spreadsheet. Generative AI is probabilistic. Ask the same question twice and you’ll get two different answers. That’s where its creativity comes from. It also has no way to know which of your versions is current. Hand it 5 and it’ll blend them, and so will every agent you run.

AI amplifies clarity. It also amplifies confusion. It chews through any language left unattended.

Software teams figured this out a long time ago.

The code is the source of truth. It changes every day, and anyone can open it and know what’s current, which is how a hundred engineers work on one product without a meeting to sync. Your company’s thinking can run the same way: written down, kept current, and read by your people and your agents alike.

Codify your thinking: treat what you mean like code.

What Your Team Keeps Bringing Back to You

Start with one decision others could make without you.

Think of the people who wait on you for your input, buy-in or approval.

For most founders and CEOs it’s a short list: which deals to chase, how much to discount, what ships next, who gets hired first. Pick the one that comes back to you most often. Or, if it’s easier, the one you made last week. Write down how you made it.

Say it’s “which new deals to prioritize”:

• Principles. How you see it. “We’d rather win five customers who’ll expand than fifty who won’t.”

• Practices. How it gets done. “Every qualified deal has an executive sponsor before the second call.”

• Conditions. What changes the answer. “We give a discount for a design partner who’ll give us a case study, and never for a logo alone.”

• Definitions. Words that mean something specific here. “Qualified means a budget, a deadline, and a sponsor.”

If your team and your AI can read that and make the call you would have made, you’ve codified one decision.

A Prompt to Codify Your First Decision

Copy the prompt below into Claude or ChatGPT and give it 15 minutes.

It’s a stripped-down version of how we work with executive clients and teams at Storied. It starts from what you’ve actually decided, then interviews you, writes the guide, and tests it. It works best in an AI that can see your email, Slack, or docs, and in the workspace your company trusts with confidential material.

Help me write down how I make one decision, so my team and my AI tools can make it the way I would, without waiting on me.

The decision: [one decision people wait on you for, e.g. approving discounts]

Not sure which one? If you can see my email or Slack, look at what people ask me to approve most often and suggest three.

Start with evidence, not my opinion. If you can see my email, Slack, docs, or notes, find the last 5 to 8 times I made this call and tell me what I actually did: the patterns, the exceptions, what I needed to see before saying yes, and any words I use in a specific way. If you can’t see any of that, ask me to paste 4 or 5 real examples, names changed, and work from those. Point out anywhere I contradicted myself.

Then interview me, one question at a time. Wait for my answer. Push back when I’m vague or when my answer doesn’t match the evidence. Stop after 8 to 12 questions, or sooner if you have enough.

Write a decision guide under 300 words, labeled v0.1 with today’s date, in four parts:

Principles: how I see this decision

Practices: how it gets done here

Conditions: what changes the answer, and what still has to come to me

Definitions: the words that mean something specific here

Test it. Give me three realistic scenarios, one an edge case. Decide each using only the guide and show your reasoning. When I correct you, update the guide, bump the version, and note what changed in one line.

Finish by formatting the guide so I can paste it into a Claude Project, a custom GPT, or our team wiki.

Run it on the decisions that most often bottleneck your team. Once your thinking is codified, people stop queuing up for your answer, your AI agents can act as your surrogate, and the next time the strategy changes you update five pages instead of holding five meetings.

Once your thinking is written down, it stops depending on who was in the meeting.

Get it out of your head. That’s how you stop being the bottleneck.

Michael

Since then I’ve done a lot of work with capital allocators across asset classes, from venture to private equity, and I’ve learned from some of the best in the game.

The most principled and disciplined of them have a way of talking and writing that signals they have nothing to prove. Concise, clean, confident. Without ego posture..

I realized I needed to level up how I communicate.

So I built an AI framework for it - which you can now download below.

Quiet Confidence: the secret to sounding like someone with nothing to prove.

I’ve watched the same dynamic with my tech and corporate clients for years. Anytime there’s power asymmetry, in role, hierarchy, or reputation, it shows up in emails, Slack messages, strategy docs, and decision memos. The writing comes off as insecure, needy, entitled, egotistical, or combative.

And almost none of it is visible to the person writing it.

The Six Hidden Power Leaks

Most of us run a few of these without ever seeing them.

1. An unclear ask. “Let me know your thoughts” gets no decision back

2. Social validation. Gushing to make the other person feel good, or look good

3. Unrequested rationale. Nobody asked why, and you told them anyway

4. Over-explaining. You make the case, then make it again

5. Hedging. “I think we could probably look at maybe” leaves them guessing

6. Apologizing. “Sorry for the delay” opens on a debt

One test catches all six. If you can cut a sentence and lose nothing but reassurance, it was a leak.

Senior people write short for two reasons: They don’t have much time and they don’t worry about proving themselves. Their work and ideas speak for themselves.

You can write this way from any seat. Done with principle rather than posturing, it signals that you want to relate to people as a peer, not as a subordinate.

Quiet Confidence Is Free on GitHub

Quiet Confidence is a voice register skill that you can download and use.

Install it inside your preferred AI, and it catches the leaks you can’t see in your own writing, whether emails, slack, decision memos, working docs, etc.

It includes a highly structured set of principles, practices, and examples.

GitHub is a software code repository. Traditionally used by open source developers, it’s now a popular AI infrastructure. Why? AI runs on documentation. GitHub is built for managing any code base, and with AI, language is the new code.

Take a look.

This is a small demo of what we’ve been building with Storied AI. It’s a set of 30+ Narrative Intelligence® skills that get your ideas funded, bought, and believed in, that you can call through an MCP connector right inside Claude and ChatGPT.

Quiet Confidence is one small example.

Three ways to use it:

• Paste prompt. No setup. Paste it into any chat, then paste your draft

• Claude skill. One file, installed once

• Custom GPT. A build kit for your own GPT in ChatGPT

Use it anywhere you’re managing up, down, or across.

Short codes for calling the skill include: “quiet confidence” “GP register”

If Quiet Confidence leads to a holy f*ck balls experience, hit reply and tell me.

And if you can, star the repo. It’ll make my mother proud.

Michael

Every company has a name for their annual planning cycle.

• Meta: the H1/H2 review.

• Amazon: runs it as OP1.

• Salesforce: it’s V2MOM.

I’ve worked on these for executive clients inside all three companies.

Different framework, same process. The template shows up now. You write the case for the thing you want. Then it moves up a chain of people who each have their own agendas to protect. Then it gets read.

Your skip level has four of these strategy docs in their inbox. Your finance partner already has a number in mind. A committee reviews it on a Thursday without you. The board eventually sees only the total.

In that file is next year’s budget, next year’s headcount, and the thing you’ve been trying to fund for three quarters. It goes in at full size and comes back at half.

Underwrite the Downside: your proposal gets cut to the size of the risk someone else has to carry. Take that loss on your own books and the upside stays full size.

Pull up what you’re planning to submit. Or if it is easier start with FY2026 strategy. How many of those sentences describe what happens if it doesn’t work?

The Cut You Make in October

The expensive version doesn’t happen in the review.

It happens first, alone, at a keyboard. You write the number you think will clear instead of the number the work needs. The second hire comes out. The scope narrows to something nobody will object to. You tell yourself you’ll come back for the rest next year. Two people on your team are waiting on one of those lines.

Nothing in the document resists it. The sentence underneath still reads fine. Nobody sees that happen. It never makes the minutes.

And it’s the version that teaches your company what this idea is worth.

Why the Number Comes Back Smaller

An ask looks like a request for money.

What it’s doing is moving exposure off your desk and onto somebody else’s.

You brought the upside. Somebody has to supply the downside, and when it isn’t in the document they build one themselves. The one they invent has no ceiling. The one you name does. Out of the last three years of things that quietly stalled. Out of what their own boss will ask in a March review. And mostly out of the number they can absorb without it becoming a story about their judgment.

Then they size your ask to fit that number.

Your document left that number blank. That isn’t caution. It’s arithmetic on an incomplete page.

The Four Lines

An underwriter prices a risk and then agrees to carry it. The second part is what lets somebody else take the upside.

1. The loss. What failure looks like, in their terms

2. The price. What it costs them, not what it costs you

3. The carrier. Who absorbs it. Say that it’s you

4. The residual. What still stands the morning after

The first three make you credible. The fourth changes the arithmetic.

A failure that leaves nothing behind is a write-off. A failure that leaves infrastructure standing is an asset, bought at a price you set in advance, and you’re the only one in that meeting who can price it.

They can accept that or not.

They can’t accept it if it isn’t on the page.

What the Document Said About Itself

This summer I created a new offer.

To build the language system a company runs on.

The prospect’s problem: their strategy was being conveyed by five different teams in five different shapes. And a market-facing message that worked with one investor customer often didn’t translate with the next.

One section in the proposal named what they could do without Storied. The raw materials are plain text MD files and a private GitHub repository. Both are commodities in this AI era. Their own team could build that code repository in an afternoon, and I said so on the same page that was asking them to invest.

The sentence after it is the one that won the deal.

The container is a commodity. Architecting language a whole organization runs on is a specialized skill and expertise. And that’s not something that one can do on their own. The failure was named before anybody asked for it.

The Price of the Experiment

Add one page to what you’re submitting.

Not a risk register. One page that says what the failed project costs, who absorbs it, and what the company still owns the morning after.

That last part’s the move. A binary ask is win or lose, and somebody staring at a coin flip buys the smallest coin on offer. An ask with a residual or compounding value is a purchase. A failed experiment can still leave infrastructure the next three projects run on, or options nobody had in January, and those keep paying.

Price that, and the number stops being what the gamble costs and becomes what the learning costs.

The number locks in December.

Name what survives the failed experiment and you can’t lose.

Michael

I left out the part that decides whether it works. Who it’s for.

Raise the Ceiling: the largest truthful version of your company is somebody else’s thesis. Nobody can build one out of a functional description, because there’s nothing in it to bet on.

Take the paragraph you had used to describe what you’re building. Could somebody write an investment case out of it?

How the Person Deciding Works

A capital allocator starts by writing a thesis.

Not a summary of your company. A claim about how the world works, one they believe most people haven’t caught up to yet.

Then a position they’re willing to take because of it. From there they go looking for the companies that make the claim true.

That’s the whole job.

A narrative is the same object. Same claim, same position, same exposure if it turns out to be wrong.

Hand over the version that’s only what it is and how it works, and you’ve handed them a product.

They can price it. They can set it next to the two others on the list and pick. What they can’t do is write a thesis. Nothing in it says anything about the world.

This isn’t only investors. The budget committee does it. So does the board review. So does the group deciding whether you’re a line item or the thing the plan gets built around.

It’s also why an initiative that matters gets funded at last year’s number. A functional description can only ask for a functional budget.

The Founder Who Sells Rails and Pitches a Wedge

One of the founders I work with builds real-time payments infrastructure. When money has to reach a person in another country the same day something has gone badly wrong for them, his system is what moves it.

He pitches it as a revenue wedge.

That’s accurate and provable. It’s also the kind of thing a buyer sets against two other vendors.

The larger true version is that he’s building the rails for insurance flows at planetary scale.

Somebody can take a position on whether the world needs those rails. Nobody takes a position on a wedge, they just decide whether it’s cheaper than the last one.

The first version gets him compared. The second gets him underwritten.

Three Things a Thesis Has

Read your paragraph against these.

• A claim about the world. Not about what you built

• A claim you could be wrong about. If nobody could disagree with it, there’s nothing to underwrite

• A position. What you’re doing about it, at the size you actually mean

Most of the paragraphs I read have the position and neither of the other two. They describe a company at a larger size without ever saying what’s true about the world that makes the larger size correct.

A description can miss all three and still read well.

That’s why so many of them survive the edit.

The Case Somebody Else Has to Make

That someone is the person writing the thesis.

I didn’t tell you that in May. Which is probably why the paragraph you wrote is still sitting in a doc nobody opened again.

You weren’t writing a description of your company at a higher altitude.

You were writing the first draft of the case somebody else has to make on your behalf, in a conversation you won’t be in, to people you’ll never meet, who will decide something about your company using words you chose months earlier.

That changes what belongs in it.

It also means you already have most of it.

This is the layer we’ve been building toward with Storied AI, so that paragraph holds its shape across every person and every model that repeats it. Readers here see it before anyone else.

Whose thesis are you making easy to write?

Michael

It’s my wife’s birthday. We’re seeing old friends, going whale watching, and catching the first NFL game ever played here.

And one big stage on Wednesday.

It’s a partner meeting at a firm whose entire craft is verifying things that already happened.

I’m there to talk about engaging people in the age of AI. How you manage and develop talent while the work gets rebuilt underneath everyone.

None of it can be verified as the workforce I’m describing doesn’t exist yet. That’s the job.

So who do you get conviction on something that doesn’t exist yet?

Disciplined Speculation: you make a bet underwritable by naming what has to be true, and what it costs to find out.

If something on your desk is asking you to commit ahead of the evidence, keep it nearby.

The Past Tense Is the Easy Part

Auditing is the cleaner half.

You look at what a company did and say whether it’s true, against evidence that already exists.

Then the same firm has to decide what it becomes 5 years from now. If the proof were sitting there it wouldn’t be a strategy, it’d be a schedule.

It’s a split that shows up almost everywhere I work.

A claim about last year gets a footnote and three levels of review. A claim about 5 years out gets a slide.

Nobody chose that. It’s what happens when an organization has a rigorous instrument for what already happened and none at all for what hasn’t. It still has to make the call on a Tuesday morning, with the same people who spent the previous hour checking someone else’s arithmetic.

The Number That Survives the Cut

The plan names the must-wins. Three of them, sometimes five.

Each one is defensible, and nobody argues.

Between now and those must-wins, nothing. No chain of reasoning, nothing anyone can retrace without the person who wrote it. Conclusions handed down, in a market that shifts while the document is still open.

Where a number is required, it gets sharper.

You take the number you believe and move it. Up if you’re asking, down if you’re promising. Then you defend the moved number as though it were the real one.

A padded forecast. Everyone reading the plan knows it happened, and nobody says in which direction.

It gets you through the meeting.

What it costs you is downstream. Nobody can tell which part of the plan is conviction and which is cushion, so the whole thing gets discounted at one rate.

Your real bet ends up priced like your padding.

The Cheapest Way to Find Out You’re Wrong

I’ve been spending a lot of time these days with capital allocators. VC, PE, corporate enterprise.

It comes down to two questions:

1. What has to be true for this thesis or investment to prove out?

2. How much does it cost to find out?

Run those against your own bet and you get three lines:

• The claim. What you’re asserting about the future, in one sentence

• The conditions. What has to be true for that claim to hold

• The price. The cheapest thing that would tell you, and when you’d know

That’s the top layer, and most of the value.

A speculative bet becomes something someone else can underwrite. Not because you proved it. Because you priced it, and you said out loud what would change your mind.

The Proof You Still Owe

I keep telling people the argument was never the problem, you can add the evidence and tighten the deck and run the revisions again. The thing still won’t move.

I believe that. It’s most of what I do for a living.

And Disciplined Speculation hands the obligation back. If language can move belief ahead of proof, you owe the proof. On a date, at a price you set yourself.

And by the way, pricing it is two moves. You price the opportunity first, because nobody funds a hedge. Then you price the downside, and what it costs to bring that down to something the company can carry.

You don’t need the evidence to make the claim.

You just need to be the one who names what would settle it and what it costs to find out. That’s available long before anyone hands you a budget.

So name the price on that thing on your desk.

What would it cost to find out you’re wrong?

Most of us try to close that gap by making the case better. More evidence, a tighter deck, extra rounds of revisions. It often never works, because the gap was never in the argument. In reality, there’s often something else going on.

Conviction Asymmetry: The side with the clearest language wins. A small team with clear language beats a large team with unclear language, and it isn’t close.

On a scale of 1-10, how clear is your strategy / ask?

How to Tell If You Have Conviction

Answer these four questions. Without any hedging.

• The future. What is the pay off, in a “must have”, visceral sort of way

• Why now. What makes this urgent to act on now, the stakes if you don’t

• The trade off. What are you willing to say no to or give up to go after it

• The rationale. What evidence do you have to support taking a big leap

Answer all four cold and you have conviction.

Now you see why conviction is the currency and often in such short supply.

One Word Can Change Your Prospects

A very large company I work with has a strategic initiative that needs funding.

One of the most important things they’ll do this decade. Yet this group has been struggling with access to next-level investment. Nobody argues against it. It just isn’t being funded at a level that matches the ambition or the needs.

An “efficiency” play.

The problem was how it was being described.

While efficiency still has its place (just ask Mark Zuckerberg in 2023), efficiency is now an AI commodified race to the bottom. People solve efficiency by just throwing AI tokens and automation at it. Efficiency is “incrementally moving the metric”. Which makes it sensible, reasonable, and nice to have. But not urgent, can’t live without.

So we’re now reframing the work from efficiency to underlying infrastructure. The data rails that ensure the company can deliver on an existential big bet. Now it’s a must-have, can’t-live-without. With a more direct path to oversized funding.

From nice-to-have to must-have.

That’s part of conviction asymmetry. The clearer size wins.

Why Conviction Gets Watered Down

Two things erode it, and both are ambient right now.

The first is decision by committee. Legal takes a pass, comms takes a pass, three stakeholders take a pass, and what comes out the other end is defensible and completely inert. Nobody degraded it on purpose. Each person removed the one part that made them nervous.

The second is vanilla beige, it sounds like everything else. Competent language is in infinite supply now and costs nothing, so the market is flooded with sentences that say nothing at all. Saying nothing is the fastest way to sound like everyone.

Clarity has an edge to it because it requires choice.

POV is the new currency.

When the Honest Answer Is No

Conviction can’t be faked or manufactured.

But it can be drawn out, distilled, and developed. It’s sourced from the inside out, which means the move is to go find whoever already holds it, usually the most senior person with the widest view of the business, and get the story out of them.

It’s one of the requirements we have at Storied.

We only work with CEOs, founders, or an executive in charge. Because a narrative has to index for conviction. And typically that’s where it concentrates.

But conviction asymmetry challenges this precept.

At a time when people don’t know which way is up, down, left or right…

Conviction fills the vacuum.

You don’t have to be “in charge” to have conviction. In fact, conviction is the greatest career-accelerating move you can work on. You just have to make sure that your conviction is legible - meaning that others can understand what’s underneath it.

Be that person.

What’s more likely, you getting fired for high conviction or for low conviction?

Michael

Margolis beach vibes

The perfect end to Summer.

Back next Sunday.

Michael

It’s now wired in how we think, how we work, and how we build narratives.

We regularly redefine expectations and raise the ceiling on what AI-generated documents can do for an organization.

The biggest hang-up with AI: when it sounds like AI.

Machine robots don’t always talk like a human. The sentences are a little slick, too staccato, or hollow and devoid of life. As a reader, you start discounting or questioning what you’re reading. Because it feels flat, trite, or just trying too hard.

What you need is an anti-AI slop file.

We’ve created one and we’ll share it with you at the end of the newsletter.

A few cliche examples

Word salad that signals lazy thinking dressed as a fancy outfit:

• “Here’s what I want to tell you.” Throat clearing

• “Stop doing X. Start doing Y.” A binary presented as a finding.

• “It’s never been easier / harder to...” Sounds deep, means nothing.

• “The real work is...” Glorifies the invisible layer, gatekeeping as wisdom.

• “Not this. Not that. But this.” Two strawmen and a reveal.

• “In a world where [X], [Y] becomes [advantage].” Urgency without substance.

• “You don’t need more X. You need Y.” Pretends to simplify while saying nothing

• “Here’s the truth nobody tells you.” Declares depth, then delivers the obvious.

You probably have your own pet peeve versions of these eye roll phrases.

When AI turns to slop

Most of AI garbage can be sorted into 3 language buckets.

1. Performative: It’s writing that tries too hard. The cloying set-up, the long expository. The cheap shortcuts. The ahem…throat clearing, barf.
   ​

2. Reductive: When AI hands back an overconfident answer. Or worse, a generic one, that makes you sound basic b*tch. Or reveals you’re talking out of your ass, and you don’t even know it.
   ​

3. Flattery: When AI fluffs you up. The sycophant. the fluffer nutter. This is the most dangerous, because our egos love encouragement and validation.

Unfortunately, you have to manually override these default settings.

The solution? An anti-AI slop file

We’ve written our own here at Storied.

And it fires every time we prompt AI. It checks our work for AI slop.

The stuff that makes me AI “vomit in my mouth.”

What we all want to read - is something with voice, personality, point of view. A deeper shade of soul. A little swerve. Some extra sass. Not vanilla beige. Or Restoration Hardware gray. Or general issue khakis.

This is why documentation matters so much in an AI world.

Many of our files at Storied now live in private GitHub repos. That’s just a fancy way of saying we’re AI nerds with a repository that lets us work with language as a code base. With a boot run file, governance and version control.

Don’t worry, you don’t need that fancy setup. To kill the AI slop.

GitHub is a great place to also share code and docs publicly.

Today we’re sharing our first one. Something you can immediately download and use.
Just point Claude or ChatGPT at the URL and ask it to install it into your LLM of choice. Either in a folder or workflow.

https://github.com/storied-inc/resources

The cliche formulas and patterns above are in there, each with an instruction for what to do instead. The rest of it is the specific tells. This is what survive a normal edit because they’re structural rather than a matter of word choice.

Other people have written their own versions, which is the point. Take ours or write your own. Just stop letting AI sound like a drunk spokesperson on your behalf.

AI is f-in incredible!

You just have to train your dragon.

Michael

Last month the biggest bottleneck in it turned out to be me.

Technically, our project management systems are Monday.com and Notion.

In reality, my project management system is a 4 x 4 grid I redraw by hand every week.

Little boxes, half of them filled in. Nearly 70% percent of the coordination across the business still bottlenecks through me. Our core team has grown back to 8 members with more than 20+ active workstreams.

Admittedly, I suck at managing the coordination layer.

I just haven’t built my way out of it yet. Even with a sophisticated AI stack.

The moment it became undeniable was a small one. One of my sharpest teammates didn’t realize that a 12-month client engagement was wrapping up that week. She hadn’t missed anything. It just wasn’t codified in the right places she would see.

I was making 500 decisions a week and not reading enough of them back to the people doing the work, even my AI agents. So the work kept getting done, and then kept getting redone, once the context finally caught up.

That gap is my own creation. And I’d guess some version of it is yours too.

The version you’re least unhappy with

You know the other shape this takes.

You sat down with an AI model to write one paragraph. You’re twelve exchanges deep now. It’s close. It keeps being close. Every pass hands you a version and none of them hands you the version, and you can’t quite say what’s off.

You had the tiger by the tail the whole time. You just couldn’t get it to hold still.

Eventually you ship the one you’re least unhappy with and move on.

This AI-first experience is near universal now, and almost nobody has named it.

Everyone is holding a different part

Now multiply that session by everyone on your team, this week, separately.

Each person has their own AI model. Each landing on their own least-unhappy version. Every one of them is right, from where they stand. Each is holding a real piece of the thing, described accurately.

It’s the parable of the 12 blind men and the elephant. One has their hand on the trunk, the other an ear, and another the leg. Every report back is honest. No two matches. And the work can’t move until someone reconciles all of them.

That’s the drag. Not the work itself. The reconciling.

I’ve started calling it Narrative Drag.

The gap between how fast you’re producing and how fast you’re actually moving.

I saw this up close for a software company whose own strategy promised customers faster decisions. Strong promise. When we opened the evidence underneath it, the proof was running on claims nobody had refreshed in a long while. Their own leader called it hope evidence. The promise had outrun the language carrying it, and every team repeating the promise was quietly repeating a slightly different one.

Back in February I described this as the Coordination Tax.

That was the right name at human speed. A tax is flat. You pay it once per handoff and move on. Drag compounds and scales with speed. The faster you produce, the more versions there are to reconcile, and every AI agent you add produces more of them.

Last week I wrote that when production is free, understanding becomes the constraint. This is what the constraint actually costs you.

Almost everyone is spending against their own velocity right now, and the spend is going downstream. Faster drafts, faster decks, faster code, faster agents. An output machine bolted onto the end of the pipe. Almost no one is working upstream, where meaning gets decided before anything gets produced. That is where velocity actually comes from, and it is sitting in plain sight with no one assigned to it.

Velocity is a language problem, not an efficiency problem.

Count the re-explanations

This stays invisible for a structural reason.

Every other source of friction in your business has an owner, a dashboard, and a budget line. Narrative drag has none of these. Nothing measures it. So no one is fixing it. Which is exactly how it becomes the largest drag coefficient.

You can see it once you know where to look.

It shows up at the re-explanation. The number of times this week you explained the same initiative to a different person, in different words, and watched them still not quite land it. That count is the size of your drag.

A large professional-services firm I’m working with is building a session for its senior people around one shift, from informing to influencing.

Informing describes the work.

Influencing drives the decision.

Activity-shaped language reports efforts. Outcome-shaped language unblocks decisions. They named the payoff in a single word. Speed.

You already have names for this. Version control. No single source of truth. Lost in translation. Coordination thrash.

Every one of them is accurate. And every one of them belongs to somebody else. So the problem gets described correctly four times and owned zero times.

Velocity is a language problem, not an efficiency problem.

The upstream fix is the thing we’ve been quietly building, an architecture that decides what things mean before anything gets produced. Storied AI is now live in early private beta. We go public in 30 days.

For now, count your re-explanations this week.

Each one left another version behind. Nothing else measures this drag. That number is the first thing that does.

Hit reply and tell me yours.

Michael

He already runs his whole story through his sales team, his prompts, and a fleet of agents building in the background. When I asked what he was most worried about, it wasn’t producing enough. It was whether all of it still said the same thing.

He’d figured out early what most teams are about to learn the hard way. When you can generate anything, the scarce thing is a story that holds its shape everywhere it shows up

So try what he did. Read ten things your company shipped this week, back to back.

Drafts, decks, agent outputs.

Do they read like they came from ten different companies?

Every team is shipping more, the priorities keep multiplying, and the org is running in many directions at once. Nobody’s sleeping on the job. You just got faster at making things than at understanding them, and the gap quietly became the work.

AI drove the cost of production to near zero. But it left the cost of understanding behind.

With Infinite Outputs, Clarity Is the Scarcest Asset

Information now multiplies faster than interpretation.

So the thing in short supply is no longer content. It’s the shared understanding that makes any output make sense..

Start with what AI actually made cheap: efficiency, the incremental 10% move. It’s already clarity-rich, that metric is obvious. And AI now hands it to you in infinite supply.

Efficiency is cheap now. Getting value, not so much.

A 10% gain is the same game played a little better, but a 10x leap is a different ballgame. There are infinite ways to play the first and only a handful to win the second. That takes the discipline to make the hard trade-offs and put everything behind the one bet that matters. And that discipline runs on the understanding AI can’t manufacture for you.

So the scarcity flips. The binding constraint is no longer output. It’s clarity: knowing which bet to make, whether the ten things your company shipped cohere, what any of the flood actually means.

The companies that pull ahead spend their scarce clarity on the one bet that matters. The rest keep pouring cheap output into 10% gains, doubling down on the one thing that was never scarce.

Shared Language, or a Faster Mess?

AI raises the stakes on all of this, and adds a new twist.

Agents are a new part of the workforce now, a new flow of work that takes understanding to manage, not more output. Point forty agents at an unmanaged language layer and they multiply the confusion at machine speed. Give them a shared one and the same volume compounds instead of scattering.

The scarcity bites hardest where the market itself is stretching. Your customers now sit at wildly different points on the curve. Take a company whose buyers span a frontier lab living years ahead, a company all-in on building, and a legacy brand still making incremental moves. One message across that spread collapses clarity. You have to name where each one sits and meet them there.

And it bites at the level of a single artifact. A nonprofit that I advise started with one pitch and ended up with six, each tuned to a different major donor. Every version read fine on its own. Together they had no center, six documents making six slightly different arguments, with no canonical one left. The donors felt the incoherence before anyone could name it.

Six versions isn’t range. It’s one story losing the thread six times, because understanding goes scarce the moment you make one thing do every job.

The Company That Can Trust Its Own Output

The prize?

A company that can trust what it ships because understanding is shared, not scarce.

The visible shift is small and it changes everything: the scarce resource isn’t output anymore. It moved to understanding, and most of your new capacity is still aimed at the side that’s already abundant.

When understanding is the scarce resource, the leverage is a shared language layer that lets your people and your agents make sense of the flood at the speed they produce it. Building that layer is what we’ve been working toward with Storied AI: an ontology for strategy and comms. More on that soon.

So run the test this issue opened with.

Read ten things your company shipped, back to back, and count the number of different companies that are described. If it’s more than one, you’ve found where your clarity is going.

Reply and tell me what you saw.

Michael

The leadership agree at the offsite. A week later, no one can convey it with conviction to their teams So it stalls at the top and never reaches its full potential. The reason is quieter than a rollout problem.

You handed everyone a set of conclusions, without the deeper reasoning.

Avoid Conclusions Without a Why

Instructions earn compliance, not conviction.

Every company runs on two systems:

1 - The visible one you manage, the org chart, the roadmap, the strategy deck.

2 - The invisible one that actually governs how people act, what they believe before they decide.

Belief architecture is that second layer. You build it by putting the beliefs, the principles, and the assumptions on paper, so a leader can retrace how you got there and rebuild it on their own.

You can’t hand someone conviction.

They earn it by retracing the reasoning for themselves. Show the trade-offs you weighed, the assumption you’re betting on, the case for the change, and your leaders arrive at the strategy instead of receiving it.

Keep the reasoning in a black box and you get nods that never become behavior.

Turn the Discipline Inward

There’s a famous firm I know whose reputation rests on showing exactly how it reaches a conclusion.

That rigor is the thing its name stands for.

Now it has to turn that same rigor inward, so its own leaders can rebuild the argument in front of a client without the deck. The firm that shows its reasoning to the world is teaching its people to show it to each other.

That is when conviction starts to scale.

The more common pattern is the opposite. An AI strategy races out ahead of where the company’s own people sit. The thinking is right. It is expensive. And it lands as an order from the top that nobody can locate themselves in. The fix isn’t a louder rollout.

It’s showing the chain of reasoning, so people can find themselves inside the strategy instead of behind it.

The Single Source of Truth To Keep You Honest

Belief architecture only works when the beliefs are actually written down.

That document is what keeps you honest, and increasingly it’s what keeps your AI honest too. Every real transformation keeps moving, and without the reasoning fixed somewhere, each cycle drifts a little further from the last.

The written layer is the North Star you steer back to.

It matters more now because AI scales whatever meaning it inherits. Point your agents at a strategy whose reasoning was never made explicit and you distribute the conclusion at machine speed with none of the belief underneath it. Agents are a new part of the workforce, and managing them takes understanding, not more output.

Belief is what produces interpretation, and interpretation is where an AI agent, or a leader, either reads your world right or reads it wrong. Anchor the beliefs and you can hold the reins loosely and let the system reason for itself. Skip the deeper work of mapping and writing them down, and you don’t just lose the narrative.

You lose the strategy, the sense of who you are, and the confidence the people around you have in the future you’re selling.

Traceability is Legibility

Take the strategy your leaders keep nodding at.

Write the one layer of reasoning you never showed them: the trade-off you weighed, the assumption you’re betting on, the case for the change. Hand them that, instead of the conclusion one more time.

Because people carry what they can retrace. They set down what they were only told.

That layer, the reasoning under the conclusions, is what we’ve been building toward with Storied AI: an ontology for strategy and comms, so it travels to your people and your agents alike. The future is almost here.

Michael

P.S. When you’re ready, here’s three ways I can help:

• If you want a second set of eyes on your narrative, reply and tell me more

• If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3

• If you need a keynote speaker, I do them on a select basis. Let’s talk

This book is where I first wrote about the SFB Narrative Method.

In case you never read it, here’s the quick Cliff Notes version:

Make hard-to-explain work easy to fund, buy, and believe in.

The more complex, disruptive, and world-changing it is, the better.

In short, it goes like this:

1. See It: Context for Change

2. Feel It: Emotional Dilemma

3. Believe It: Evidence of Truth

This is what makes a narrative most compelling and hard to resist.

Treat SFB as a series of decision gates that must be followed in a precise order.

The problem is that most of the time, we lead with Believe It. Forcing data and conclusions on an audience that isn’t primed or enrolled. Then we wonder why we hit so much pushback. Often it’s because they don’t get the big picture (see it), and they don’t feel it’s for them (feel it). In my experience, this is why 90% of pitches fail.

Agreement and readiness are different states.

A case built to win the argument often never builds the sense that saying Yes is the only responsible thing to do.

Belief Is Inevitability, Not Agreement

A room commits when you make the decision feel inevitable.

Not louder. Inevitable.

That happens via three decision gates, in sequence.

• See It: makes the shift undeniable. You zoom out and give shape to what’s already happening, so the room shares your read of reality before you ask for anything.

• Feel It: makes the stakes irresistible. You zoom in on what standing still actually costs, so the decision matters instead of merely being correct.

• Believe It: makes the yes irrefutable. You price in the risk with proof and a staged path, so acting becomes the safe move and hesitating becomes the expensive one.

Undeniable, irresistible, irrefutable.

Run in that order, they build the conditions a Yes comes from.

The Blank Check Nobody Argued For

A little while back, I worked with an exec who needed to pitch their CTO on an ambitious AI bet.

Big tech company you’d definitely know. This was when budgets were being cut, layoffs were rampant, and my exec’s org was wildly misunderstood and undervalued.

Yet despite the odds stacked against them, my exec client exceeded expectations. They got everything they asked for and more. And the pre-read did the whole job.

The CTO didn’t even want to have a meeting about it.

Not “should we fund this,” but “how much headcount can you hire, and how fast?”

My exec was shocked how they were just handed a blank check.

That’s the moment you dream of. The one where someone just hands you their wallet and says, “take my money.” Because staying put just became the riskier bet.

Anyone who allocates capital, like CFOs and investors, thinks this way.

They hold a thesis, price the risk, and model the upside. Then they stage the commitment against a series of gates. A decision narrative does the same. It leads the room to where the downside is capped, the upside is too big to ignore, and saying no is the larger mispriced move.

The Layer AI Can’t Automate for You

Your AI can draft the deck, the memo, and the model.

It can generate every version of the argument you’ll ever need.

But it can’t make a room ready to decide.

The output layer is now a commodity. The decision layer is the one still worth your name on it. See It, Feel It, Believe It is how you run it. It’s the difference between producing a case and building the readiness to act on one.

Strip the setting and the move holds. A philanthropic venture fund just put a structural bet in front of a top investor in their own language, a concentrated position on a mispriced problem, staged against gates so the first commitment de-risks the next.

Board vote, investor wire, budget commit. Different contexts, same sequence. Undeniable, irresistible, irrefutable.

The diagnostic falls out of it too. A stalled decision isn’t a verdict on your case, it’s a read on which of the three gates is thin. If the room is lost, it can’t see it. If it’s indifferent, it can’t feel it. If it’s hedging, it can’t yet believe it.

You fix the blocked decision-gate, and you have a winning narrative.

The Audience That Decides With You

There’s a version of this where the audience doesn’t just agree with you.

It decides with you.

The next time you carry a decision you need won, take 5 minutes before the meeting and name which of the three decision gates your case is actually missing. Not the slide that needs polish. The gate. Then solve for that one.

Because the goal was never a better argument. It’s the moment the YES stops feeling like a risk and starts feeling inevitable.

That whole sequence is teachable, once AI actually knows your worldview, your stakes, and your evidence. It can run it with you instead of just drafting words. That’s what we’ve been building with Storied AI: an ontology for strategy and comms.

The Claude MCP server just went LIVE this week.

This is getting real. Fast.

Michael

P.S. If you haven’t read Story 10x yet, you can download the PDF here for free. Or find it on Amazon in all the formats: hardcover, audiobook, kindle. And yes, it also wired into Storied AI as a set of Claude skills.

P.S.S. When you’re ready, here’s three ways I can help:

• If you want a second set of eyes on your narrative, reply and tell me more

• If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3

• If you need a keynote speaker, I do them on a select basis. Let’s talk

But in 2004, I lived in Washington DC and briefly dated a tango instructor. It’s how I was introduced to the culture of milongas, a late night social tango party.

At the time a new sound was making its way onto the dance floor. Electronica-beats mixed with classical tango. A wild fusion of world music, trip-hop, dub step, and beyond.

Nobody had a name for it.

I was early in my narrative practice, and Sharna, an old friend, was trying to describe this rising cultural phenomenon. One day, over coffee, we gave it a name: Neotango.

Sharna wrote a blog post on it, and it quickly took off from dancers to DJs to festival flyers. Somewhere in there it stopped being Sharna’s word and became the category’s. Two decades later, it’s just what this music is called.

For an introduction to the genre, check out Gotan Project or this longer list.

Whoever names it, owns it.

That’s the topic of this week’s newsletter.

Picture a leader who hears their own words describing their category in a place they were never present. A grant application. A competitor’s pitch deck. A regulator’s language. An AI answer written for someone three companies away. The words they almost never bothered to write down are the ones the field now uses by default.

Getting there is rare, because it cuts against how the game is usually played. You’re differentiating inside a category someone else named, so every pitch answers a competitor’s frame. You can out-think your whole market and still be renting a position in a house another company built. That’s a share game, and a share game caps you.

Last week we talked about giving your narrative a long enough half-life to compound. This is what compounding buys you once it runs long enough.

Bigger Than a Bigger Slice

The most valuable position in any category isn’t market share. It’s owning the language the field uses to describe itself.

When your words become the default way people name the problem, the opportunity, and the stakes, everyone else is drafting off you. That’s a Semantic Monopoly. A market-making move. You’ve defined the category everyone else competes inside.

You build it the slow way. You codify a small set of ownable phrases, the ones that say here’s what we do and here’s what this whole space is, and you repeat them until they stop sounding like yours and start sounding like everyone’s.

There’s a discipline underneath it: knowing which language to spread and which to protect. The category language, the name and the thesis for the whole space, you give away on purpose. The more freely it travels, the bigger the category you anchor, and every company that adopts it is extending your position for you, usually without noticing. Your proprietary language, the codified frameworks and coined terms that are distinctly yours, you protect, because that’s the asset everyone else drafts from.

There’s a third layer between those two, the shared argument a coalition of partners carries in their own voices, and it’s where most ecosystem work quietly breaks. That one’s a longer conversation. If you’re living in it, reply and tell me.

The Machines Inherit Whoever Named It

AI raises the stakes on all of this.

Models scale whatever language they inherit. So the terms your field uses become the terms its AI repeats, at volume, across every output. Own the language and you own what the whole category generates from. Leave it unowned and the machines default to someone else’s frame, or to the flat average of everyone’s.

This is the rare position AI can’t erode. The models in your category are already running on somebody’s words.

The only question is whose.

Strip the Names and the Move Repeats

A coalition in a new category of frontier tech, with real regulatory stakes, reduced a sprawling mission to one named problem. Their funders realized the named problem was the actual product. A wide range of players who would never fully agree could suddenly pull the same direction, because they’d all inherited the same language.

A global life-sciences company wrote the words for how breakthroughs clear regulatory approval. The industry adopted them. The framing traveled all the way up to the regulators themselves. Now they’re the source everyone else drafts from.

You already know the finished version. Salesforce named “the cloud,” and an industry reorganized around the word. Palantir built around the word “ontology” for data, and for that world and use case, the word is now theirs.

And it’s continuous work. A membership organization I’m working with is building the shared language for its whole field right now. It’s the hardest part of the job, because coherence across a network isn’t everyone saying the same words, it’s everyone making the same argument in their own voice.

The One Phrase You’d Make a Rival Use

The move takes 5 minutes.

Name the one phrase you’d want your whole category to adopt.

The words you’d want a competitor forced to reach for in their next deck. Then start saying it, everywhere, in every memo and talk and prompt, until it turns into wallpaper.

Does your language travel without you in the room? That’s the tell: the words showing up where you never were, carried by people you never briefed.

Someone’s language becomes the default your category thinks in. The only question is whether it’s yours.

That default now has to hold across every agent your company runs, not just the people who work there. There’s a language layer a category thinks with, both humans and machines. That’s what we’ve been building toward with Storied AI: an ontology for strategy and comms, not just data. More on that soon.

Michael

P.S. Sharna never trademarked neotango. That’s exactly why it won.

P.S.S. When you’re ready, here’s three ways I can help:

1. If you want a second set of eyes on your narrative, reply and tell me more

2. If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3​​

3. If you need a keynote speaker, I do them on a select basis. Let’s talk

On Saturday I took my girls (age 2.5 and 1) to their first LA Dodgers game.

And it had me reflecting on this 250th birthday of America.

While it feels like there are few things we can agree to in U.S. politics right now...

I can sense a new narrative taking shape. Especially as we head into the mid-terms this fall.

The reason its hard to see is that a truly positive story of America doesn’t get told enough these days.

And you have to repeat an idea or message consistently for it to compound.

That’s the topic of this week’s newsletter.

The Sharper Version That Keeps Costing You

Recently, a chief of staff told me they rewrote their company’s website once a week for 16 weeks straight.

Every version landed. Their CEO would read it and say this is great, this is exactly it. The clarity held for about 36 hours. Then a sharper way to say it would surface, a new frame, a better paradigm, and the whole thing started over.

They weren’t failing. They were doing exactly what their CEO asked. And nothing they wrote ever got the chance to stick, because the walls kept moving before anyone downstream could build on them.

This is the trap for the sharpest leaders. The instinct that built the company, keep finding a better version, is the same one that keeps resetting the clock.

The Version Worth Leaving Alone

A message compounds the way anything does.

It needs time and repetition. The team internalizes it, new hires inherit it, the market hears it often enough to say it back to you. Every rewrite sets that clock to zero.

So the discipline isn’t finding the best possible words. It’s finding the version durable enough to leave alone. Narrative half-life: how long your core narrative survives the rate of change before it genuinely needs a rebuild. A short half-life means you’re re-articulating constantly and nothing accrues. A long one means the framing holds while everything around it moves.

The work is separating what’s load-bearing from what isn’t. One Series C SaaS company had a CEO years ahead of their own org on where the category was going, taking a legacy product AI-first. They kept re-articulating the vision, and every re-articulation quietly restarted the company. The fix had nothing to do with the vision. It was naming the one version that was load-bearing, committing to it, and letting the team experiment at the next layer down while the core held still.

The company started compounding behind them instead of chasing them.

A Generic Voice You Keep Rewriting to Escape

Your AI runs on your language now, too.

Point a model at your message and, left alone, it drifts toward the generic. Anthropic says it plainly about its own model: it has a consistent default house style that is persistent. The critic Kyle Chayka gave the result a name, the generic style of AI web design. That flat, could-be-any-company voice you keep rewriting to escape is the machine averaging you toward the middle.

You don’t fix that with another rewrite. You fix it by giving the model a durable layer to run on. Anthropic’s own analytics team found their AI answered business questions correctly about 21% of the time on its own. Once they gave it the codified context or ontology, accuracy ran past 95%. The model didn’t get smarter. It got a stable language layer to stand on.

Now multiply that across a company. A growth stage tech platform pointed a fleet of agents at a message that changed every quarter, and got 40 versions of itself running at machine speed, each one a slightly different company. Because AI amplifies clarity. It also amplifies confusion, and it will chew through whatever language you leave unattended.

One durable narrative is the only thing that scales cleanly across the people and the machines both.

Leave the Good One Alone

A sharper version will always show up.

When it does, the question isn’t whether it’s better. It’s whether the one you’ve got has had long enough to compound. Name the one version of your narrative that’s load-bearing, and give it a quarter before you touch it. Let it travel. Does your language travel without you in the room? That’s the test that actually counts.

This is the layer we’ve been building toward with Storied AI: the narrative work I do with clients, now callable tools inside your own Claude, working on your real drafts, the raise, the board update, the all-hands. Not to spin up more versions, but to find the one durable enough to compound and hold it across every agent you run.
​
Want to early beta access? Hit reply.

Michael

P.S. Baseball, hot dogs, cracker jacks. That’s one story we can still all agree on.

P.S.S. When you’re ready, here’s three ways I can help:

1. If you want a second set of eyes on your narrative, reply and tell me more

2. If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3​​

3. If you need a keynote speaker, I do them on a select basis. Let’s talk​

The next chapter is obvious. The market shifted; the old way of winning is dying.

But when they stand up at the all-hands to lay it out, it doesn’t land. The people who built the company over the last 10 years hear something else: the thing they poured a decade into creating is now what’s holding everyone back.

The harder a CEO pushes, the harder a team digs in.

That’s the real problem with change. Not the strategy. It’s that the people living in today can’t yet see the tomorrow that’s obvious to you. The people whose buy-in you need most are often standing inside the very thing you’re trying to change. From where they sit, the new story isn’t an exciting new future. It’s a verdict on the past that they’re still attached to.

Thankfully, there’s one move that can fix this. And it fits on a single slide.

Old Story / New Story

You put where the company has been alongside where it’s going.

Two columns, side by side.

Not to argue that the new way is better, but to let people see the shift at a glance and connect the dots themselves.

For that CEO, it looked like this:

• Product → Platform

• Feature → Foundation

• Bought once → Built on daily

Three lines, and their whole bet was legible.

The company that sold a product is becoming the platform other companies build on. You can feel how that lands differently than 10 slides of strategy. The bigger the story, the more room there is for people to find themselves inside it.

The best pairs stick because they’re built for: parallelism, a little alliteration, a rhyme you can’t unhear. Feature to foundation. People don’t just read them, they repeat them.

And whoever sets the frames that stick owns the conversation.

Why people stop fighting it

The frame works because it never makes the old way wrong.

It names the old as where you honestly were, the product that earned the company the right to become a platform, and the new as where it was always heading. Same conviction, next chapter. Nobody has to have been wrong for everyone to move. Take that away, and people reject the new story the way a body rejects a transplant it didn’t ask for.

Leave it in, and the thing they built becomes the foundation of the thing they’re building next.

It’s already happening, in worlds nothing like yours

A team I worked with inside a life-sciences company had spent years as the function everyone treated as the brake, the people who slowed the science down to keep it safe. They didn’t argue for more respect. They redrew the map.

• Gatekeepers → Enablers

• Risk management → Opportunity management

• Following the rules → Rewriting the rules

Same standards, new altitude. The brake became the hand on the wheel: someone is going to write the protocols for how AI enters their business function, and it should be the people who know a thing or two about compliance, regulators, and quality.

A strategy leader inside a large software company did the same with a data bet the company kept shelving. Inside the building it read as overhead. So they stopped pitching a fix and went back to the company’s origin story, the data model that let it scale in the first place.

• Overhead → Infrastructure

• Cost center → Growth engine

• Agents run the data → Data runs the agents

They named the asset, “ownable data,” and what read as a cost started reading as the company’s next platform. Same DNA, carried forward. End result? A blank check from the CTO with unlimited engineering headcount.

Build your own in 10 minutes

Take a piece of paper, and draw a line down the middle: two columns.

On the left: where you’ve been, in your people’s own words, the version they’ve lived. On the right, where you’re going, one line for each line on the left, each pair built to echo the one beside it. Three rows carry most of it: the promise, the shift, and the one thing that never changes. Then read it back to someone still living the old story, and watch their face. When they nod, you’ve built the bridge. The change that needed a deck now fits on one slide.

One VP I taught this to called the technique, “worth its weight in gold.” They now include a contrast frame in every business case and slide presentation. And their decision-making influence has grown alongside it.

Why people start moving toward it

This move doesn’t just simplify the change. It pulls people into it.

The hardest thing about any transformation is that the people standing in today can’t see the tomorrow you’re describing, let alone want it. A strong contrast frame closes that gap. It shows people who they get to become and the future they get to inhabit. A future people can see themselves in is almost impossible to reject. You stop dragging people toward the change, and they start walking toward it on their own.

For years this lived in my client workshops and advisory work. And this is one of the first skills going into Storied AI, a Claude plug-in that you can call on in the flow of work. All you have to do is describe what’s changing, and it creates the frames that win with investors, executives, and B2B buyers.

I’m now releasing the first narrative stack for tech operators, and opening it to a small group to pressure-test before it opens to everyone. It’s free for this first group, and I’m keeping it small so I can work with each tester.

Want in? Reply “Storied AI” and tell me the change you’re trying to land.

Michael

P.S. When you’re ready, here’s three ways I can help:

1. If you want a second set of eyes on your narrative, reply and tell me more

2. If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3​​

3. If you need a keynote speaker, I do them on a select basis. Let’s talk

Their positioning deck, sales decks, marketing content, with a note on top:

“I hate most of this. Don’t use any of it yet.”

I have a version of this conversation almost weekly.

They’re not alone, and it’s not their fault. Things are moving faster than most companies can bring back into alignment, so the words end up a step or three behind the company’s strategy and GTM. Hate it or not, your document library is the language your whole company runs on. Every output your team is producing is likely pulling from a corpus that’s gone stale and no longer fits where you actually need to go.

So, why does language often lag behind the company you’re already become?

You’re running on legacy language

Generic is only the surface tell.

There’s a word for the technology a company has outgrown but still runs on.

The word is legacy. Legacy processes, legacy systems, legacy infrastructure. It once served a purpose, and now it quietly taxes everything you try to build on top of it: the simple change that somehow takes three engineering sprints, the workaround nobody remembers the reason for. We call it tech debt, and right now everyone is pointing AI at it, ripping it out and rebuilding.

Almost nobody realizes they’re carrying legacy language too. The “averaging” I’ve written about these past two weeks, where AI reaches for the middle because you gave it no center, is one symptom of it. Outrunning your own words is the deeper one.

Words for the company you used to be are still running the company you’ve become.

How a phrase becomes yours

Ownable language is the handful of phrases people repeat on your behalf.

It’s the line a customer says back to you in a meeting you weren’t in. You don’t get it by writing a clever tagline. It comes from two things:

The first is discernment. Of all the true things you could say about what you do, which few words actually carry it? Choosing those, and cutting the rest, is taste, and it’s the one move AI can’t make for you, because it reaches for the average and the average is the opposite of a choice. Compress the complicated thing down to that essence, and the phrase starts to read less like a description and more like a rallying cry.

The second is repetition. A sharp line said once is just a tagline. You come to own it by living it: using it, repeating it, building around it until other people say it back to you. That’s the difference between a trademark and a tattoo. A trademark is registered. A tattoo is worn, and it’s costly to remove because you made a real commitment to it.

Take a life sciences company we work with. Their strategy lived in a document many pages long that no one could repeat from memory. We worked to give them a few words clear enough that the whole company, product and sales alike, could rally around the same idea.

That idea started with the clinician. Their days fill up with the work around the work: administering tests, writing reports, deciding which assessment to use at all. So the company said it back to them in their own language: “Your work is hard enough. Let’s make it easier.” And the promise underneath it: “Give people their time back.” The proof sits beneath it: hours of friction pulled out of the day, and time to diagnosis cut by 30 to 50%. Empathy for the clinician has become the through-line for the product, the experience, and the way the company talks about where it’s going.

What those words actually do

Signature words do more than travel. They bind.

When a company lands on a few words in common, people stop drifting into private versions of the strategy. A new hire in week one reaches for the same line a 5-year veteran uses. That shared language is what culture actually is, not the values poster on the wall, but the words everyone reaches for without being told.

Get a company onto the same handful of words and it’s finally singing from one song sheet.

That used to be true only of your people. Now it has to hold for your AI, the agents and models your company is starting to run on. They inherit whatever language you hand them, so the more distilled that language is, the more it behaves like code.

Codify it once and it aligns the team and every agent at the same time. Words as code and words as words. Distilled language has gravity; it pulls people and AI into the same orbit. Leave it generic and it goes inert, nothing repeats it and nothing coheres around it, and your language quietly commoditizes into the category, the opposite of a moat.

This is why voice and point of view aren’t decoration on top of strategy.

With AI, language is both a currency and the building material, the instructions your company runs on and the meaning those instructions carry, at once. The work is to chisel it with enough distinction that it’s worth repeating and impossible to mistake for anyone else’s.

Naming a category that didn’t exist yet

A real-estate investment firm we work with isn’t selling buildings.

It’s selling a thesis: that longevity, the wellness and recovery economy, is an asset class in its own right, not an amenity you add on, and that culturally relevant places hold their value. The language carries the firm: longevity as an asset class, ownership in places that matter, the shift from transactional to transformational real estate. Those phrases didn’t exist in the category before this firm earned them. Now they’re how it describes a new kind of firm for a new kind of time, and how the market is starting to describe it too.

Your best line is probably already in your deck

A B2B startup we know spent months trying to frame its category.

The line that finally did it was already in their deck, buried on slide 12, in their own words: “the trust layer for B2B.”

It had been there the whole time. They just had to move it to the front and commit to it.

Sometimes the words are already in the room, in how a founder talks off-script or a phrase a customer keeps using. Sometimes they have to be found, framed, and shaped. Either way it’s its own discipline, which is why it helps to have someone who obsesses over language and meaning, whether that’s a narrative architect or someone on your team who owns it. And it matters more now, not less.

AI is a large language model. The more you organize your language into a codex, the more precisely you can scale it, across every person and every agent that draws from it.

It’s also what we’ve been building toward at Storied: a way to codify this language so it holds across your whole team and every AI agent, not just the one LLM chat window you’re in right now. Next week, I’ll start showing you some of Claude skills that can do this for you.

So, a question to leave you with: what’s the one line you’d want people repeating about you? And are you saying it enough times that they will?

Michael

P.S. Language is the new code. ;)

P.S.S. When you’re ready, here’s three ways I can help:

1. If you want a second set of eyes on your narrative, reply and tell me more

2. If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3​​

3. If you need a keynote speaker, I do them on a select basis. Let’s talk