Take the work apart, see it, then decide who does each part.

Part 1 Mark’s account

The org chart is a poor map of the work

Org charts show who reports to whom. They don’t show how an enquiry becomes an invoice, who really approves what, or where work waits. In the organisations Mark has worked with, nobody held the whole picture.

Part 2 Proposal

Every piece of work has the same shape

Something comes in, something is done to it, something goes out. Describe work that way and it becomes comparable: you can ask who or what does each piece best, a person, software, AI or a mix.

Part 3 Proposal

A capability has four properties

A capability is a piece of work with a clear input, transformation and output, and someone who answers for it. To count, it should have:

1. A clear interface: what goes in and what comes out, at any level. There is no smallest size.
2. Measurable performance: how well and how much.
3. A person who answers for it, whoever or whatever does the work.
4. Persistence: it keeps its identity as the people or tools inside it change, like a football club whose players change.

Part 4 Hypothesis

Capabilities are made of capabilities

A team is made of people; a company of teams; a network of companies. If each level is described the same way, the whole organisation could be seen as one map: Market → Find → Engage → Sell → Deliver → Support, rather than CEO → Director → Manager.

Part 5 Proposal

Trust needs a record, failures included

To choose a capability you need its record. We propose every claim is marked Verified (checked from records or by someone else), Declared (the owner says so) or Projected (expected, not yet shown). And the record must carry failures: a record that can only carry good news isn’t evidence. It’s marketing.

The Periodic Table of Work puts this into practice →

Part 6 Proposal

Delegate the doing, keep the deciding

However good software and AI become, an organisation must keep five things: its intent, authority over what it commits to, accountability, ownership of what makes it distinct, and the ability to change its mind. AI can act on authority it’s been lent, and that authority can be taken back. It can’t be the root.

Part 7 Counterargument

How this could fail

• Coordination costs: stitching many small capabilities together can cost more than it saves.
• Know-how nobody wrote down: much valuable work is hard to describe as inputs and outputs.
• Gamed metrics: measure “fifty leads” and you may get fifty poor ones (Goodhart’s Law).
• Lost apprenticeships: if small tasks go to software, how do people learn the job?

We’ll treat these as tests, not footnotes.

Where it leads

From twin to digital corporation

Put these ideas in order and they make a journey: build a twin of the organisation, take its work apart into capabilities, reassign them, and put them back together. We’re mapping the stages in public. See the working model →

Sources

Where the ideas come from

We’re not the first to take jobs apart. Work Without Jobs (Ravin Jesuthasan and John Boudreau, 2022) already argues for breaking work into tasks and putting it back together. Our claim is narrow: the join of a twin built from an organisation’s own systems, one card for people, software and AI, search by input and output, a record with failures, and a named person behind each. We’ll report what we find, either way. The library lists the work we build on.