Context engineering is the new management skill

When an AI agent joins your team, the question changes. It isn't "how do I motivate this contributor?" It's "what does this contributor know, and does it know the things it needs to?" The answer is almost never yes by default. The organisation's context - policies, conventions, failures from last year, the unwritten rules - has to be made accessible.

Three kinds of context

I find it useful to split context into three buckets. Factual context - what is true about our systems and data. Normative context - what we've decided to do, and why. Tacit context - the patterns we've learned that nobody has written down.

Factual context is the easy one: documentation, schemas, runbooks. Normative context is harder, because organisations often don't remember why they decided things. Tacit context is hardest of all, because by definition it exists in the heads of the most experienced people.

The manager of the future is not just a motivator of humans. They are a curator of the context their organisation operates in.

What changes for leaders

You stop writing one-off docs and start designing the system through which context is captured and surfaced. You stop thinking of onboarding as a human-only activity. You start caring, explicitly, about which decisions are legible and which aren't - because the legible ones are the ones that scale.

The leaders who will look like magicians in the next decade aren't the ones with the best prompts. They're the ones whose organisations have the clearest, most accessible context surfaces - so every contributor, human or otherwise, can do their best work.