Are You Ready for AI Agents? A 20-Point Readiness Checklist

AI agents can research accounts, qualify enquiries, answer customer questions, draft follow-ups and update systems. The technology is increasingly capable. The harder question is whether your business is ready to use it safely and usefully.

An AI agent is an AI system that can take actions, such as looking up data, updating a record or triggering a workflow, within limits you define. That ability to act is what makes agents valuable, and what makes preparation essential.

Use the checklist below. Answer each point honestly with yes or no.

A. The use case

1. We have chosen a specific workflow, not a general ambition to use AI.
2. The workflow happens often enough for improvement to matter.
3. We can describe what a good outcome looks like for that workflow.
4. We know what the work costs today in time, delay or missed opportunities.

B. The data

5. The information the agent needs is written down and up to date: policies, product details, processes.
6. Our CRM or core system data for this workflow is reasonably accurate.
7. We know which data the agent should not be able to see.
8. We have the right consent and legal basis to use customer data this way.

C. The process

9. The current process is documented, including common exceptions.
10. We have removed unnecessary steps rather than planning to automate them.
11. It is clear which systems the agent must read from and write to.
12. The people affected understand why the change is happening.

D. The guardrails

13. We have decided which actions the agent may take alone and which need approval.
14. There is a clear hand-off to a person when the agent is unsure or the case is sensitive.
15. Agent actions will be logged so we can review what happened and why.
16. We have a set of real examples to test the agent against before launch.

E. Ownership and measurement

17. One named person owns the agent’s performance.
18. We have a baseline measure to compare against.
19. There is a regular review of quality, errors and customer feedback.
20. We have a plan to switch the agent off or roll back if needed.

How to read your score

• 16 to 20 yes answers: you are ready for a focused pilot with human review.
• 10 to 15: promising, but close the gaps first. Data and guardrails are the usual culprits.
• Fewer than 10: start with foundations: process documentation, data quality and ownership. An agent built now would likely disappoint.

What a sensible first agent looks like

Good first agents share a few traits. They work on high-volume, low-risk tasks. They read more than they write. Their output is reviewed by a person at first. And their success can be measured simply, for example time saved per enquiry or share of routine questions answered correctly.

Examples include summarising an account before a sales call, drafting replies to common support questions from approved content, or reading inbound enquiries and suggesting a qualification tier.

Where human judgement stays

Pricing decisions, contractual commitments, complaints, sensitive personal circumstances and anything irreversible should remain with people, supported by AI rather than replaced by it. Designing that boundary deliberately is what makes teams trust the system.

Next steps

If your score shows gaps, you are in good company; most organisations have them. Our AI agents and assistants work starts with exactly these questions. For a broader view of your readiness, take the Growth Systems Assessment.