The nine dimensions of human-agent readiness

Readiness for AI agents is a property of the whole system of work. Sociotek looks at nine dimensions: six about the organisation that must absorb the change, three about the conditions agents need to operate safely.

Warm marks the human system: judgement, relationships, accountabilityCool marks the agent system: identity, permissions, execution
1

Strategic leadership alignment

Whether the people at the top agree on what AI is for here, and on which responsibilities may be handed to an agent and which must stay with a person who answers for them.

Ask yourself: If each executive wrote down what an agent may decide alone, would the lists match?

2

Workflow and value readiness

Whether anyone has looked hard at how the work really flows before automating it, and chosen what to drop, simplify, assist, automate or keep human.

Ask yourself: Which steps in this workflow should disappear before any agent touches them?

3

Work, role and decision design

Whether it is clear who owns each agent, what it may decide, and how the jobs of the people around it have been reshaped.

Ask yourself: When the agent gets it wrong, who answers for it by name?

4

Trust, participation and employee experience

Whether the people doing the work were told the truth, had a say in the redesign, can tell when an agent is acting, and are treated fairly when it errs.

Ask yourself: Who among the people doing this work helped redesign it?

5

Learning and adaptive capability

Whether people are learning to brief, check and overrule agents, can report mistakes safely, and are still building the judgement the business will need.

Ask yourself: Where will your juniors learn what the agent now does for them?

6

Change and scaling capacity

Whether managers can lead the change, whether there are clear tests for scaling or stopping, and whether oversight grows as agents multiply.

Ask yourself: What evidence would make you stop, and who can call it?

7

Agent infrastructure, identity and orchestration

Whether each agent is known to your systems as itself, limited to what its task needs, and able to hand work on without losing the record of who was allowed to do what.

Ask yourself: Could you list every agent at work today and what each may touch?

8

Governance and responsible autonomy

Whether accountability for AI is clear from the board to the front line, and whether each agent's freedom is set by how serious and reversible its actions are.

Ask yourself: Which actions should an agent never be the only one to check?

9

Operational assurance and resilience

Whether you can trace what an agent did, stop it quickly, and undo what matters.

Ask yourself: If an agent went wrong at scale this afternoon, how long until you noticed, stopped it and put it right?

The dimensions work together. A high score on infrastructure does not make up for people who were never consulted, and the reverse is also true. How the diagnostic uses them