An AI readiness checklist is a structured set of questions that shows whether a company has what it needs to use AI usefully and safely: a clear business problem, accessible and reliable data, the right systems, people who can own the work, and rules for risk, privacy and the EU AI Act. A company is ready when it can name one specific use case, point to the data behind it, say who will own it and explain how success will be measured. This guide gives the checklist in seven areas, explains how to score it, who should own it and how often to review it.
Last updated: 13 November 2026.
Is my company ready for AI?
Your company is ready to start with AI when you can answer yes to four questions: Do we have a specific problem where AI could save time, reduce errors or improve service? Do we have the data that problem depends on, in a usable form? Is there a person with the authority and time to own the project? Do we know how we will judge whether it worked?
Readiness is not a single state. Most companies are ready for some uses, such as summarising documents or drafting replies, long before they are ready for others, such as automated decisions about customers. The checklist below helps you see where you stand for each idea.
At Odduel, a creative and technology agency in Barcelona and São Paulo, we use a version of this checklist at the start of projects in our AI development service. It complements our guide on how to start an AI project, which covers the first project itself.
In short: check strategy, data, systems, people, governance, security and measurement. Score each area honestly, fix the gaps that block your first use case, and start small. Readiness grows with each project, so review the checklist every six to twelve months.
What should an AI readiness checklist cover?
A useful checklist covers seven areas. Work through each with the people who know the facts, not only with management.
1. Strategy and use cases
- We have a short list of specific problems where AI might help, written in business terms.
- Each idea has a named owner and a rough estimate of the value if it works.
- We know which ideas are low risk (internal drafting, search) and which are high risk (decisions about people).
- We have chosen one first use case that is valuable, feasible and low risk.
- Leadership supports a small pilot with a clear end date.
2. Data
- We know where the data for our first use case lives and who owns it.
- The data is reasonably complete, current and consistent.
- We can access it through exports or interfaces, not only by copying from screens.
- We know which data contains personal information and on what legal basis we hold it.
- Documents have a current version that is clearly marked.
3. Systems and technology
- Our main systems (CRM, ERP, help desk, document storage) offer interfaces for integration.
- We know which tools we already pay for include AI features.
- We have cloud accounts or infrastructure that can host a pilot securely.
- We can run a test without touching live customer systems.
4. People and skills
- Someone in the business can own the use case day to day.
- Staff who will use the tool are involved in testing it.
- We have a plan for basic AI training, as the EU AI Act requires providers and deployers to take measures to ensure AI literacy among their staff.
- We know which skills we have internally and which we will bring in.
5. Governance and compliance
- We have a simple policy on which AI tools staff may use and what data they may enter.
- We know whether our use case falls into a high-risk category under the EU AI Act.
- Our data protection officer or adviser has reviewed the use of personal data.
- Contracts with AI providers state where data is processed and that our data is not used for training.
- There is a human check before AI output affects customers or employees.
6. Security
- Access to AI tools follows our normal sign-in and permission rules.
- Sensitive documents are excluded or protected.
- We log usage in a way that respects privacy.
- We have considered risks specific to AI, such as prompt injection and leaked data in outputs.
7. Measurement
- We have measured the current process: time taken, error rate, cost or satisfaction.
- We have defined what success looks like for the pilot.
- We will collect user feedback during the pilot.
- We have a decision point: scale, change or stop.
How do you score the checklist?
Score each area from one to three:
| Score | Meaning | What to do |
|---|---|---|
| 1 | Not in place: blocks any use case | Fix before starting a pilot |
| 2 | Partly in place: enough for a low-risk pilot | Start small, improve in parallel |
| 3 | In place and documented | Ready for wider use |
You do not need threes everywhere. A low-risk internal pilot can start with twos in most areas, as long as governance and security are not at one. Higher-risk uses need stronger scores across the board.
The scores matter less than the conversations they start. If data owners and managers score the same area very differently, that gap is the first thing to resolve.
What are the steps to use the checklist?
- Gather a small group: a business owner, someone from IT, someone who knows the data and a person responsible for data protection.
- List three to five candidate use cases in plain language.
- Go through the checklist for the most promising one, scoring each area.
- Mark blockers: anything scored one in data, governance or security.
- Agree actions for each blocker, with owners and dates.
- Choose the first pilot once blockers are cleared.
- Repeat for the next use case, reusing what you have already fixed.
Many companies find their first useful pilot is an internal assistant over company documents; our guide to an AI internal knowledge base explains how that works.
Who should own AI readiness inside the company?
Ownership works best when it is shared but clear. A member of the leadership team should sponsor AI work and make decisions on priorities and risk. A business owner should be responsible for each use case and its results. IT should own security, integration and tool choices. The data protection officer or adviser should review personal data use. In a small company these may be two or three people; what matters is that each role has a name.
Avoid making AI readiness a purely technical project. The questions that most often stop a pilot are about data ownership, process and permission, not models.
How often should it be reviewed?
Review the checklist every six to twelve months, and whenever something important changes: a new system, a new use case, a change in regulation or a new AI provider. The EU AI Act applies in stages, so obligations relevant to your uses may change over time; check the official text and its timeline, and ask your legal adviser when in doubt.
Each review should be quicker than the last, because each project leaves behind cleaner data, clearer rules and more experienced people.
Which frameworks can help?
You do not need a formal framework to start, but recognised references help larger companies structure their work. ISO/IEC 42001 specifies requirements for an AI management system, covering policies, risk and continual improvement. The EU AI Act itself sets out the obligations by risk level. Use these as references for the governance part of your checklist, scaled to your size.
Questions people ask
What is an AI readiness checklist?
It is a list of questions that shows whether a company has what it needs to use AI usefully and safely: a clear use case, accessible data, suitable systems, people to own the work, rules for risk and privacy, security measures and a way to measure results. Scoring each area shows which gaps to fix first.
Is my company ready for AI?
You are ready to start when you can name one specific problem AI could help with, point to the data it needs, name a person who will own it and explain how you will measure success. Readiness differs by use case: low-risk internal uses can start well before automated decisions about customers.
Who should own AI readiness inside the company?
A leadership sponsor sets priorities and accepts risk, a business owner is responsible for each use case, IT owns security and integration, and a data protection officer or adviser reviews personal data. In small companies two or three people may share these roles, as long as each role has a clear name.
How often should an AI readiness checklist be reviewed?
Every six to twelve months, and whenever something important changes, such as a new system, a new use case, a new AI provider or a change in regulation. The EU AI Act applies in stages, so obligations can change. Each review becomes quicker as data, rules and skills improve.
Does the EU AI Act affect small companies using AI tools?
Yes, in proportion to how they use AI. Companies deploying AI systems have duties such as ensuring staff have sufficient AI literacy, and stricter obligations apply to high-risk uses. Most everyday internal uses carry lighter duties. Check the official text and ask your legal adviser how it applies to you.
Pick your most promising AI idea and run it through the seven areas this week: the gaps you find are your plan. If you want help assessing readiness or building a first pilot, see our AI development service or get in touch.


