Enterprise AI Readiness: A Practical Assessment Framework

Most enterprise AI initiatives I'm asked to look at have already decided what they're going to build before anyone asked whether the organisation is ready to build it. The proof-of-concept goes well, the production rollout stalls, and someone calls me in to work out why.

Nine times out of ten the answer is the same: a readiness gap that should have been caught at the start. The framework below is what I use to catch it before the budget gets committed.

The five pillars

A useful readiness assessment looks at five areas. Most organisations are strong in two or three and weak in the others. The weak ones are where AI projects fail.

1. Data readiness

This is the one that derails the most projects. AI systems need data that is accessible, accurate, and well-described. Three questions cut through most of the noise:

• If I asked a new employee to find the definitive source for customer X, how long would it take them?
• How confident are you that two different teams pulling the same metric would get the same number?
• For the systems you'd want an AI agent to read from, who owns the schema and how often does it change?

If the answer to the first is "ask Jane, she knows", and the answer to the second is "we'd have to check", you have a data maturity problem that no AI project will work around. Fix it first.

2. Infrastructure readiness

This is the easiest pillar to assess and usually the least blocking. Cloud-based AI services have eliminated most of the heavy infrastructure lift. The questions that actually matter:

• Do you have a sanctioned path for using AI APIs (Azure OpenAI, Anthropic, Google) that meets your security and compliance requirements?
• Can your applications make outbound API calls to AI providers, or is everything locked down to private networks only?
• If you needed to put a private model behind your firewall, do you have the GPU access to do it?

For most enterprises in 2026 the answer to the first is "yes via Azure OpenAI" and that's enough. The other two only matter for specific scenarios.

3. Skills and team readiness

The team you need for an AI project is not the team you need for a typical software project. Specifically:

• Someone who can do prompt engineering well — this is a real skill and it takes a few months to develop
• Someone who can evaluate AI outputs systematically rather than just spot-checking
• Someone who understands the cost and latency implications of model choices
• A product owner who can make trade-offs between AI accuracy, cost, and user experience

You don't need PhDs. You do need people who have built and shipped AI features before, or who are given the time and budget to learn. Throwing AI projects at teams that have only ever built CRUD applications usually goes badly.

4. Governance readiness

AI introduces categories of risk that traditional software doesn't. The readiness questions:

• Is there a documented policy for what kinds of data can be sent to which AI providers?
• Do you have a process for assessing AI-related risks (bias, hallucination, prompt injection) for any new system?
• If an AI system makes a decision a customer challenges, who's accountable and how do you investigate?
• How do you know if a model update from your provider has changed your system's behaviour?

Most large enterprises have at least started on this. Most mid-sized ones haven't. Either way, you want governance answers in place before you put AI in front of customers — not after the first incident.

5. Use case readiness

This is the one I weight highest. AI applied to the wrong problem fails for reasons that have nothing to do with the technology. A well-chosen use case has:

• A clear baseline you're trying to improve — current process time, current error rate, current cost per transaction
• A user population that will actually use it (this is harder than it sounds)
• Tolerance for occasional mistakes — AI is rarely the right choice for processes that require 100% accuracy with no human review
• Enough volume that the cost of building the AI solution makes sense against the cost of doing it the old way

The single best filter is the baseline. If you can't measure what the AI is supposed to improve, you'll never know if it worked.

How I run the assessment

For a typical engagement, I spend two to three weeks on the assessment phase. Roughly:

• Week 1 — Interviews with stakeholders across business, IT, data, and security. Document review (data dictionaries, policies, architecture diagrams). One workshop to surface candidate use cases.
• Week 2 — Deep dive on the top two or three use cases. Pull together baseline metrics. Talk to the people who'd actually use the AI system.
• Week 3 — Write up findings, present, agree on a path forward.

The output is a report with two parts: a readiness score across the five pillars with specific gaps identified, and a recommended sequencing for which use cases to attempt first based on readiness and value.

What "ready" actually looks like

You don't need to be strong in all five pillars to start. You need to be honest about where you're weak and have a plan to address it. The organisations that get the most out of AI investments share three traits:

• They picked one or two use cases first and got them right, rather than launching ten in parallel
• They invested in the boring infrastructure work — data quality, governance, evaluation tooling — before they invested in flashy capabilities
• They had executive sponsorship that survived the first failure

The last one matters more than people think. The first AI project rarely lands as well as expected. The organisations that abandon AI after one disappointment never get the compounding value. The ones that treat it as a multi-year capability build are the ones that pull ahead.

Where most people land

AI readiness has less to do with how modern your tech stack is than with whether your data is clean, your skills are real, your governance is in place, and your first use cases are well-chosen. Most enterprises I assess are stronger on the tech side than they are on those four. The fix is usually rebalancing the investment, not adding more tooling.

If you'd like help running a readiness assessment for your own organisation, or want a second opinion on where you currently sit, get in touch.