The management autopilot

Companies spend a little on alignment tools and a fortune on alignment work. The OKR tracker is a rounding error next to the people who chase the OKRs: the chiefs of staff, the strategy and PMO teams, the ops leads, the EOS implementers, the integration consultants — and the tax inside every manager's week spent collecting status, reconciling roadmaps, and re-explaining the strategy. Sequoia's Julien Bek puts the general pattern bluntly: for every dollar businesses spend on software, about six go to services. The tool budget is small. The work budget is enormous. Every alignment product ever sold has competed for the wrong one.

Copilots and autopilots

Bek draws the line that matters: a copilot sells the tool; an autopilot sells the work. QuickBooks is ten thousand dollars a year and the accountant is a hundred and twenty thousand — “the next legendary company will just close the books.” Software engineering crossed this threshold first, because writing code is mostly what he calls intelligence work: complex, but rule-based. The same crossing is coming to every profession — including the one nobody thinks of as a profession, because it's everywhere: management.

What “autopilot” actually means

Aviation settled this argument decades ago. An autopilot flies the plane; the pilot remains in command. Autopilot never meant no pilot — it meant the pilot stops hand-flying and starts supervising. That is precisely the shape of the thing we're describing. The management autopilot runs the operating rhythm: it watches how work actually happens, keeps the written record reconciled with reality, rolls up goal state, notices drift, and delivers context to the person who needs it. The human ratifies what matters. Nothing enters the source of truth without approval. The ratification gate isn't a limitation on the autopilot. It's the yoke.

Management splits, like everything else

Bek's frame divides work into intelligence — rule-based, automatable now — and judgement: experience, taste, the call about what to do. Management splits cleanly along that line. A huge share of a manager's week is intelligence work: collecting status, reconciling the cascade, chasing check-ins, assembling the update, noticing that the plan and the reality have drifted apart. The protected core is judgement: what the strategy should be, who to bet on, when to pivot. The management autopilot takes the first category. The second stays human — and gated.

Why every previous attempt rotted

Companies have tried to systematize management forever: the wiki, the OKR rollout, the operating system installed by an outside implementer. These work — as long as a human pays the maintenance tax. Then the tax comes due every week, nobody can afford it forever, and the system rots into a graveyard nobody trusts. The failure was never the writing down. It was the upkeep. What's new is that the upkeep is intelligence work — exactly the kind of work AI can now do, with a human approving every change. The curation maintains itself. The human stays in command.

The compounding loop

Here's the part that compounds. Bek observes that today's judgement becomes tomorrow's intelligence as systems accumulate data about what good decisions look like. A ratification gate produces exactly that data, continuously: every proposed update, and every human approve or dismiss, is a recorded example of how a well-run company decides. An autopilot built this way doesn't just do the toil — it learns the judgement, at whatever pace its owners ratify. That's a moat no index-everything system can copy, because indexing has no gate to learn from.

Verification, manufactured

The service categories being automated first — insurance, claims, tax, medical coding — share a trait: standardized, verifiable work products. Management has no billing codes. So the work product has to be manufactured: a definitional layer where every artifact has an owner, a provenance, a verified date, and an audit trail of every change proposed and every change approved. That record is what lets an autopilot be accountable for the outcome — a company that stays aligned — instead of merely helpful. Accountability is the whole difference between a tool and a service.

What stays human

Everything that deserves to. Strategy is a human call. Hiring is a human call. The pivot is a human call. And one line is permanent: the autopilot flies the company, not scorecards on individuals. Structural misalignment — where the plan and the reality have drifted — is the system's business. Individual performance is not, and never will be.

Bek predicts the next trillion-dollar company will be “a software company masquerading as a services firm.” For fifty years, management software has sold tools and left the work on your desk. The interesting company sells the work: a company that stays aligned, delivered as a service, supervised by the humans in command. Management on autopilot. Judgement on the stick.