AI Adoption Readiness Checklist

AI adoption should begin with a clear understanding of the business problem, not with the technology itself.

Before introducing an AI tool, workflow or application, an organisation should understand its current processes, data, people, systems and objectives. Work through the sections below before starting an AI initiative. A section with several unanswered questions is usually where the initiative will stall.

01 — Business need

• Have we identified a specific business problem?
• Can the problem be described clearly?
• Is the problem recurring or significant enough to justify improvement?
• Do we understand the current process?
• Have we identified the people affected by the problem?

02 — Process readiness

• Is the existing workflow documented?
• Are repetitive activities clearly identified?
• Have we identified the manual steps?
• Have we identified the bottlenecks?
• Do we know which decisions require human involvement?

03 — Data readiness

• Do we have access to the data required?
• Is the data relevant to the problem?
• Is the data reasonably structured?
• Are data ownership and access understood?
• Are privacy and security considerations understood?

04 — Technology readiness

• Do our existing systems expose usable data or interfaces?
• Can the proposed solution integrate with those systems?
• Do we understand our current technology environment?
• Have the integration requirements been identified?

05 — People and operations

• Is someone responsible for the initiative?
• Have the affected teams been identified?
• Do employees understand how the proposed system will affect their workflow?
• Is training required?
• Is there a process for reviewing the outputs the system produces?

06 — Risk and governance

• Have privacy considerations been reviewed?
• Have security risks been considered?
• Have human-review requirements been identified?
• Are acceptable-use boundaries defined?
• Is there a process for monitoring the system once it is running?

07 — Implementation

• Is there a clearly defined first use case?
• Can the initiative be tested on a limited scale?
• Are success criteria defined?
• Is there an implementation owner?
• Is there a plan for evaluating the pilot?

---

Are we solving the right problem before choosing the technology?

Related reading: what generative AI is and where it fails, and how to start using AI without starting from scratch.