Buying the Model Isn't the Same as Being Ready

Part of what Coyote & Quill is for: we watch the frontier so you don’t have to, then tell you what actually matters.

Here’s one worth 28 minutes. Greg Brockman — OpenAI’s president — sat down with a room of founders. Watch it here.

Why it’s worth your time: the man whose company sells the model spends most of the hour telling founders the model isn’t the hard part. That candor, from that seat, is rare — and it points straight at where AI projects actually succeed or fail.

Here’s what we took from it, and why it’s exactly how we think about readiness at C&Q.

The model is one-eighth of the job

Brockman keeps circling one idea without naming it: readiness is not one thing you buy. It’s a stack of things you have to get right at once.

The military has a checklist for exactly this. Before it fields a new capability, it doesn’t just ask “did we buy the kit?” It runs down eight dimensions — doctrine, organization, training, materiel, leadership, personnel, facilities, policy. The kit — materiel — is one of the eight. Buy the shiny new system and neglect the other seven, and you are not ready. You just spent money.

The acronym is DOTMLPF-P. Ugly; the idea is clean — the equipment is the easy part; readiness lives in everything around it. Listen to Brockman again with the eight in mind:

• Doctrine — he requires “a human to be accountable for all code that gets merged.” Do you have a written rule for who signs off on AI work, or is it vibes?
• Organization — OpenAI is “still figuring it out” on how to structure teams around agents. Who owns AI enablement at your shop?
• Training — “it’s very different to hear AI described versus to use it.” Are your people trained to operate it, not just fear it?
• Materiel — the model, the tools. The one box most people think is the whole test. It’s one-eighth.
• Leadership — “human attention is going to be this incredibly scarce resource.” Do your leaders know where to spend theirs?
• Personnel — the room asks whether we’ll have “human software engineers in a decade.” Which of your roles shift from doing to deciding?
• Facilities — data provenance, permissions, the plumbing. When the AI touches a document, can you trace what it did?
• Policy — “in cybersecurity and in biosecurity, those are areas where we’re very thoughtful.” In regulated work, the human sign-off isn’t a courtesy. It’s the law.

Brockman, whose job is to sell you dimension four, spent the hour on the other seven.

Why this is how we work

This is the exact framework Coyote & Quill runs an AI-readiness assessment on: we check all eight before you pour money into the one. The failure mode isn’t buying the wrong model — it’s buying the right model into an organization that was never ready to hold it.

So when the president of the leading AI lab spends an hour making that case to a room of founders, we take it as confirmation, not news. It’s the thing we assess for — and staying close to what the frontier is actually doing is how we keep that assessment honest.

You don’t get ready by shopping. You get ready by checking the other seven boxes.

Want your own workflow checked against these eight — or a look at the plain-language checklist I’m building from them? Start a conversation.

---

Source: Greg Brockman, fireside chat (AI Ascent 2026), YouTube. Human accountability ~10:00, “still figuring it out” ~11:39, “scarce resource” ~16:22, cyber/biosecurity ~20:45.