Becoming Legible to AI

Most leaders spent the holidays either burying their heads in the sand or doom-scrolling AI announcements with a case of Sunday scaries about the ground shifting underneath them.

A few came back ready to win, focused on becoming legible to AI.

Can an AI agent actually understand how your business works? Does it know what counts as a qualified lead, which edge cases matter, and where the authoritative answer lives when finance and sales disagree?

This is the bottleneck. Despite billions in investment, only 6% of organizations are AI high performers achieving significant value, with nearly two-thirds remaining stuck still just experimenting. The AI models are capable and ready, most companies just haven’t done the work to get the context of how their businesses actually run available to those models.

Will Manidis wrote about companies becoming "legible to capital." The more legible they become to pools of capital, the more dollars form behind them. What’s hard for most companies becomes incredibly easy for them.

Same is true here. Some businesses are highly legible to AI. New models are released, and the next day these businesses are already achieving an order of magnitude more productivity. Their teams are on podcasts talking about how they’ve scaled revenue 5x without adding headcount. And their results are real.

So what does it take to be legible to AI?

1. Business logic is explicit. Edge cases documented. Rules written down. The answer isn't "ask Sarah."
2. Context is connected. Data from your CRM, ERP, internal tools, and spreadsheets are able to be combined. If each system is an island, the agent is blind.
3. Knowledge compounds. Every decision is encoded and becomes available to the next person and the next agent.

None of this is actually about AI. These are the same things that make organizations work for humans. Clear documentation. Connected systems. Institutional memory that doesn't walk out the door. AI just raises the stakes.

Companies that invested in operational rigor are discovering they’ve already built the right foundation for AI. Just ask people like Nick at Ro, Dani at Fabletics, Josh at Magic Spoon, or Marie at Whoop.

Companies that run on tribal knowledge are discovering the AI transformation they want has a prerequisite they haven't met.

Vendors are promising AI benefits everywhere. But if they're not giving you surface area to encode how your business works, they can't deliver. It’s like a brilliant employee who never onboards into your way of working. (Foundation Capital calls this missing layer the "context graph.")

AI can do remarkable things today. It can already change the entire operating model of your business. Even more remarkable things are coming this year. But you’ll only benefit if you're legible.

The companies that figure this out won't just use AI. They'll compound with it.