Home/Knowledge Base/Knowledge Worker Playbooks/K-02

K-02K-Series · Foundation

The Knowledge Worker AI Framework

If you memorise one thing about working with AI, make it this framework. It fits in two lists, it applies to every task and every tool, and it doesn’t date when the software changes. AI helps you draft, summarise, analyse and create. You provide judgement, empathy, relationships, ethics and decisions. Everything else in this series is this framework applied to real work.

Why this matters

Most AI advice fails in one of two directions: breathless lists of things AI can supposedly do alone, or fearful lists of reasons to avoid it. Both miss how good work actually happens — as a collaboration with a clear division of labour. A framework gives you that division in advance, so you’re not renegotiating it under deadline on every task.

The two lists

AI helps you:

• Draft — first versions of emails, reports, plans and papers
• Summarise — long documents, meetings and threads into what matters
• Analyse — compare, find patterns, spot gaps and contradictions
• Create — options, ideas, structures and starting points

You provide:

• Judgement — is this true, relevant and good enough?
• Empathy — how will this land with the person reading it?
• Relationships — the trust the work travels on
• Ethics — should we, not just can we
• Decisions — what actually happens next, and who’s accountable

Notice the shape. AI’s list is production. Yours is meaning. And yours is longer — that’s not an accident, it’s the argument. You bring more to the table. That’s the point.

Human, then AI, then Human

Every AI-assisted task worth doing runs in the same order: you frame the work (the purpose, the sources, what good looks like), AI contributes (the draft, the summary, the analysis), and you complete the work (verify, adjust, decide, own it). The bookends are always yours.

Almost every AI failure you’ll ever see is a missing bookend. No framing produces confident nonsense on the wrong task. No completion sends confident nonsense to your boss. The middle was never the risky part.

Using the framework day to day

Before any task, ask two questions: which parts of this are on AI’s list, and which are on mine? Then hold the line. Hand the drafting over freely; never hand over the deciding. Teams that adopt this language find something useful happens — ‘that’s a your-list item’ becomes a normal, non-awkward thing to say in a meeting.

Putting it into practice

1. Write the two lists somewhere you’ll see them for a fortnight
2. Before your next AI-assisted task, name which list each part belongs to
3. Do the bookends deliberately: frame first, verify and decide at the end
4. Catch one moment this week where you nearly handed over a your-list item
5. Teach the framework to one colleague — explaining it locks it in