Startup Announces Pivot to Whatever You Just Said

The founder of Agreeable Labs, Inc. has a gift, and the gift is this: he can hear the end of your sentence and begin his company again before you've finished exhaling.

I watched him do it in March. A partner at a fund — one of the good ones, the kind with a podcast — said the words "the interesting thing is really the infrastructure layer" somewhere around minute nineteen of a conversation about something else entirely. He said it idly. He said it the way you'd mention weather.

By minute twenty-two, Agreeable Labs was an infrastructure company.

Not considering becoming. Not exploring a wedge into. Was. Had always been, in the retroactive way these things work, the way the previous eleven months of engineering effort reorganized themselves in real time into a story that had been heading here the whole while, obviously, if you knew how to read it.

The releases, in order¶

What follows is the complete public communications history of Agreeable Labs across eleven months. I have reproduced them faithfully, which is easy, because I wrote all of them. If you've been through this desk's earnings call, you know the arrangement: the document is invented, the grammar is field-collected.

---

JANUARY — Agreeable Labs Emerges From Stealth

Agreeable Labs today announced $18M in Series B funding to build the operating system for small business logistics. "Nobody has solved the last mile for the long tail," said the founder. "We're going to."

---

APRIL — Agreeable Labs Announces Strategic Evolution

Following deep customer discovery, Agreeable Labs is sharpening its focus to the infrastructure layer beneath small business logistics. "The last mile was always downstream of the real problem," said the founder. "We've gone up the stack to where the leverage is."

---

AUGUST — Agreeable Labs Announces Agreeable Intelligence

Agreeable Labs today unveiled Agreeable Intelligence, an AI-native platform. "We've been building toward this since day one," said the founder. "The infrastructure work was the foundation the models needed."

---

NOVEMBER — A Note From Our Founder

As we enter our next chapter, we're returning to our roots: serving the small business operator directly. Some difficult decisions accompany this focus. I want to thank the 40 team members departing us for their contributions.

---

Four theses. Eleven months. Not one of them survived contact with a person holding a checkbook who happened to be thinking about something else.

The doctrine that makes it possible¶

Here's where I have to be careful, because the real thing underneath this is a genuinely serious argument and I don't want to pretend otherwise.

In September 2024, Paul Graham published an essay called Founder Mode, built around a talk Brian Chesky gave about running Airbnb. The argument: founders are handed a playbook written for professional managers, and the playbook is wrong for them. Chesky, Graham writes, was told to "hire good people and give them room to do their jobs." And then, flatly: "He followed this advice and the results were disastrous."

Graham's diagnosis is sharp:

“

what they were being told was how to run a company you hadn't founded — how to run a company if you're merely a professional manager

”

Manager mode, in his description, means "you treat subtrees of the org chart as black boxes. You tell your direct reports what to do, and it's up to them to figure out how." Founder mode breaks that — it will, he notes, "break the principle that the CEO should engage with the company only via his or her direct reports."

I think this is substantially right. Founders do get handed a costume that fits someone else. The essay is describing something real, and Graham says plainly that founder mode is still largely undocumented.

But an undocumented doctrine that licenses a founder to override the org chart is an extremely useful thing to be holding when someone asks you why the company means something different than it did in April. Founder mode has a shadow, and the shadow is that "I saw something nobody else could see" is indistinguishable, from the inside, from "a man with a fund said a word and I liked how it sounded."

Both feel like conviction. Only one of them is.

What the data says about where this ends¶

Strip out the comedy and look at the mortality tables.

CB Insights examined 431 VC-backed companies that shut down since 2023, categorizing 385 of them by cause. The results:

CB Insights is explicit that the headline number is misleading: running out of capital "is almost always the final cause of death, not the root problem." Companies don't die of an empty bank account any more than people die of a stopped heart. The question is what stopped it.

And 43% poor product-market fit is the answer, sitting right there — which is precisely the thing a serial pivot is designed to look like it's solving while actually deferring. Each new thesis resets the clock on ever having to demonstrate fit. You cannot fail to find product-market fit for a product you keep replacing.

Meanwhile, the gravitational field everyone is pivoting into is measurable. FactSet counted the term "AI" on 331 earnings calls among S&P 500 companies in Q4 — 68% of the 485 calls held. The five-year average is 149. The ten-year average is 94.

Q4 (most recent)331 calls

5-year average149 calls

10-year average94 calls

By sector, Information Technology hit 94%, Financials 91%, Communication Services 89%. At those rates it stops being information. When nearly every company says the word, the word cannot distinguish between them, which means it isn't a signal about the companies — it's a signal about what the audience is currently paying for.

That's the thing Agreeable Labs understood better than its investors did. It just said it four times instead of once.

The part where I stop laughing¶

Pivots are real, necessary, and frequently correct. Some of the most valuable companies alive are the third idea a team had. Changing your mind in response to evidence isn't weakness; it's the entire job, and any satire that lands on founders shouldn't pivot has missed by a wide margin and hit the wrong person on the way past.

So let me be exact about who this is aimed at.

A pivot costs the founder a slide and costs everyone else a year.

When the thesis moves, the founder's equity moves with it — same cap table, same board seat, same story with a new middle. The investor's position is unchanged; a fund holds twenty of these and needs one to work, and a portfolio company that keeps pivoting toward whatever is fundable is behaving exactly as the incentive structure requests. Nobody in that room is harmed by April becoming November.

The engineer who moved cities for the logistics thesis is harmed. The person who turned down another offer in February is harmed. The forty people in that last press release, who appear in the sentence as a quantity rather than as names, are harmed — and they're the only participants who had no access to the meeting where the company became something else, which is a species of the same problem as every decision that gets made in a hallway and ratified in a room.

That's the asymmetry, and it's why the joke points where it points. The absurdity isn't that companies change direction. It's that we've built a financing system where changing direction is nearly free for the people who choose it and expensive for everyone who has to follow, and then we describe the result using the vocabulary of vision.

If you are taking money, know what the options actually are before you assume venture is the only one, and read what that particular money costs before you take it — not the dilution, which everyone models, but the narrative obligation, which almost nobody does. A raise is a promise to keep being fundable. That promise has a direction, and it is not always the same direction as your customers.

How to tell a real pivot from a narrative pivot¶

Both announce themselves identically. They diverge in the details nobody puts in the press release.

1. 1

   Ask what evidence arrived

   A real pivot has a discovery attached: a cohort that didn't retain, a segment that pulled the product somewhere unexpected, a number that went the wrong way for three months. Ask what specifically was learned and when. A narrative pivot answers with a market, not a measurement.

2. 2

   Check who was in the room

   Real pivots usually surface from people touching customers — support, sales, the engineer who reads the logs. Narrative pivots arrive from a conversation with capital and travel downward. Ask where it originated. The answer is rarely hidden; it's just rarely asked for.

3. 3

   Watch the verb tense

   "We've been building toward this since day one" is the tell. A genuine change of mind can say we were wrong — it has the standing to. A narrative pivot must retroactively make the past coherent, because admitting a dead end would mean admitting the eighteen months were spent.

4. 4

   Count the pivots per funding event

   One pivot per raise, arriving shortly after the raise, is the pattern. Plot the dates. If every thesis change lands within a few weeks of a term sheet or a board meeting, the driver isn't customers.

5. 5

   If you work there, price your own exposure

   You are an undiversified investor in one company with one asset — your time — and unlike the fund, you cannot hold twenty. That's not a reason to leave. It's a reason to know your number: how many more pivots you'll fund with your own year before the answer is no.

Minute twenty-two¶

I keep coming back to those three minutes in March — the gap between a partner saying infrastructure layer while thinking about something else, and a company having always been an infrastructure company.

Three minutes is not enough time to evaluate a market. It's barely enough to finish a thought. But it is precisely enough time to notice which way the money is facing and turn to face the same way, and the terrible thing is that this is not stupidity. It's the single most rational response available to a founder whose next eighteen months depend on someone in that room feeling good on the drive home.

Agreeable Labs isn't a parody of a badly run company. It's a parody of a company responding correctly to its actual incentives — which is the only kind of parody worth writing, and the reason the fake press releases were so easy to draft. I didn't have to exaggerate the language. I just had to put four of them next to each other, which is a thing no press release ever has to survive.

The founder still has the gift. Somebody is going to say something to him next quarter.

It's just business.

Sources

1. Paul Graham — 'Founder Mode' (September 2024)
   +How this was checked

   The essay drawn from Brian Chesky's Airbnb talk: 'hire good people and give them room to do their jobs', 'He followed this advice and the results were disastrous', and the manager-mode/founder-mode distinction.

2. CB Insights — 'Why Startups Fail: Top Reasons'
   +How this was checked

   431 VC-backed companies that shut down since 2023, 385 categorized: ran out of capital 70%, poor product-market fit 43%, bad timing 29%, unsustainable unit economics 19%.

3. FactSet Insight — 'More Than 65% of S&P 500 Earnings Calls for Q4 Cited AI'
   +How this was checked

   331 of 485 calls (68%) cited 'AI'; 5-year average 149, 10-year average 94; Information Technology 94%, Financials 91%, Communication Services 89%.