The Accountability Asymmetry: Why Agents Buy Small and Humans Still Buy Big

A founder in Lena Waters’ office hours let a sentence slip that is worth more than most governance frameworks: “You can’t go to your board and say, my agent told me we should do this.”

That is not a capability statement. It is a constitutional one. It draws a line through every buying decision your company will face this year, and the line is not where most revenue leaders think it is.

The line is not drawn by smarts

The easy story is that agents handle the small stuff because they are not smart enough for the big stuff. Wrong story. Agents already compare vendors, read docs, test APIs, and issue recommendations that would embarrass a junior analyst if the analyst could compete on speed. On raw cognitive horsepower for narrow evaluation, the agent wins.

What the agent cannot do is stand in a boardroom. It cannot absorb blame. It cannot be fired. It cannot look a CFO in the eye and explain why the contract was signed. The line between agent-decided and human-decided is not drawn by capability. It is drawn by who faces the consequences.

This is the accountability asymmetry, and it is doing more work in shaping GTM right now than any model release.

What the asymmetry buys you, and what it costs

For low-stakes purchases (seat licenses, utilities, swap-outs, anything reversible inside a quarter), the asymmetry collapses. The human does not need to explain that decision to anyone. So the agent gets the keys. A growing share of those transactions are already happening without a human touching the buying flow at all. Some companies have, as Waters’ recap notes, replaced their websites with plain markdown optimized for machine parsing. They are not being clever. They are being honest about who reads them.

For enterprise deals, the asymmetry holds. Not because the agent could not evaluate the vendor, but because no one will sign a million-dollar commitment they cannot narrate upward. The boardroom is the only room in the building where “my agent told me” is still a sentence that ends a career.

Most GTM playbooks do not acknowledge this split. They treat AI readiness as a uniform problem: make the site agent-friendly, structure the data, ship the API. That is half the answer. The other half is recognizing that your enterprise motion is now aimed at a human who needs a story to tell a board, and your mid-market motion is aimed at a machine that needs a schema to parse. Those are different products of the same website.

The quiet rebuke

Waters has another line from the same session that deserves quoting in full: “Removing human coordination overhead and calling it transformation? That’s debt repayment.”

This is a quiet rebuke to every AI pitch deck claiming transformation through headcount reduction. Cutting coordination cost is not transformation. It is paying back the interest on organizational debt your company accrued when it grew faster than its processes. Useful work. Necessary work. Not transformation.

Real transformation, in the agentic era, is drawing the accountability boundary deliberately — deciding which decisions you delegate to agents because the cost of being wrong is survivable, and which decisions you keep human because someone has to face the board. Most companies have not even started that conversation. They are still arguing about which model to license.

What to do on Monday

If you run revenue, the practical move is to stop treating your funnel as one audience. You have two.

The first audience is an agent. It wants structured data, predictable APIs, machine-readable pricing, documentation that does not lie. Your job is to equip it the way you would equip an internal advocate — give it what it needs to recommend you confidently to whatever principal it serves. Visual hierarchy, emotional copy, and navigation UX are invisible to this audience. Applied human psychology, as Waters put it, does not work on something that has no psychology.

The second audience is still a human, and that human is carrying a very specific burden: they need language that survives contact with a board. Their fear is not that your product is bad. Their fear is that they cannot explain the decision if it goes sideways. Your enterprise content has exactly one job, and it is to hand them that language. Case studies that sound like boardroom narratives. Risk framings that a CFO can repeat verbatim. Governance artifacts that make the decision defensible.

If your site does only one of these, you are leaving half the market on the floor. If it does both, badly, you are in the majority. The companies pulling ahead are the ones that noticed the line, accepted the asymmetry, and stopped pretending one message could serve both sides of it.

The boundary is not moving because agents got smarter. It is moving because accountability did not. That is the governance decision hiding inside every AI GTM conversation this year, and it is the one most teams are still avoiding. See also our companion piece on selling to AI agents as a governance problem for the seller-side mechanics.

---

This essay synthesizes Tom Tunguz’s recap of Lena Waters’ office hours (Founders, Equip Your Agents, April 2026). The accountability-asymmetry framing — that the boundary between agent-decided and human-led is drawn by who faces the board, not by capability — is the governance principle we think will define B2B GTM for the next eighteen months.

Victorino Group helps revenue teams draw the accountability boundary between agent-decided and human-led. Let’s talk.