You answer “are we ready for agentic AI?” one process at a time, not once for the whole organisation. Find the steps where an agent pays off first, check that each has a named owner, a current process and a clear decision boundary, then fix what the next step needs. Readiness is a route you plan, not a verdict you wait for.

Start with what you already hold

The board pack asks for an agentic AI plan by the end of the quarter. Someone in the room asks the obvious question, and the answers go round the table: the data team is confident, IT is cautious, operations wants to know which work is in scope.

Each of them is right about their own area. That is the useful finding.

You already hold most of what the answer needs. Process models in your repository. Procedures and work instructions. Risk registers and the controls your auditors test every year. And the knowledge your experienced people carry about how each process runs and where it bends. The work is not to start again. It is to see how far you can trust each of these, and for which processes.

Ask the question in a form that has an answer

“Are we ready?” invites a yes or a no, and neither helps you plan. A better question is: where will agents pay off first, and what does that step need to succeed?

Readiness has several parts. TDWI’s Agentic AI Readiness Assessment (TDWI, January 2026) scores it across five dimensions: organisational readiness; data and context; technology and engineering; governance, risk and control; and operationalisation and learning. Technology is one of five. An organisation can be strong on platforms and still unsure who owns the processes an agent would touch.

So separate two decisions:

  • Organisation readiness. Ownership, governance, skills and data across the business. This sets how quickly you can scale.
  • Process suitability. Whether one process, with its steps, decisions and controls, fits an agent today. This sets where you start.

Treating them as one question is why the room gives five answers. Treating them as two gives you a plan. For the second decision, see how to tell whether a process suits an AI agent.

Four things to know about any candidate process

When you look at a process as a candidate for an agent, four facts decide most of the answer.

  1. A named owner. One person accountable for the process today, not a team or a department.
  2. A current process. A documented version that still matches how the work runs, with a date and a source you can point to.
  3. Clear decision boundaries. You can say which steps carry a decision, who is accountable for each, and the limit within which it can be made.
  4. A way to recover. You can see what was done, stop it and put it right.

Where all four hold, you have a strong first candidate. Where one is missing, you have a specific action, and usually a quicker one than rebuilding a pilot later. Setting the limits in point three is its own discipline; we cover it in how to set safe autonomy for AI agents, step by step.

Where IGX fits

We work from the knowledge you already hold. The first step is seeing how far you can trust it: who owns each process, how current it is, where two sources disagree and where the evidence runs thin. That view turns “are we ready?” into a short list of places to start and a clear set of things to strengthen.

It also changes the board conversation. Instead of a confidence level for the whole organisation, you can show which processes are ready for an agent now, which need one or two actions first, and which are better left human-led for good reasons. That is a plan a board can fund.

Being clear that some work should stay with people is part of the answer, not a weakness in it. A well-run human-led process is a sound decision when variation, relationships or consequences call for judgement.

What to do this month

  • Ask each function the same questions. Operations, IT, risk and the executive team often answer differently. The differences show you where shared understanding needs work.
  • Shortlist three processes. Choose ones with high volume, a named owner and decisions you can describe. Check each against the four facts above.
  • Turn every gap into an owned action. Give each one an outcome, an owner and the evidence that will show it is done. We set out how in turning agentic AI readiness gaps into owned actions.

The organisations that move fastest with agents are the ones that know where to start, and the first project you deliver well is the one that makes the rest possible.