How to assess artificial intelligence readiness

Artificial intelligence readiness means being prepared to adopt and operate AI within a defined business process, with evidence for the data, ownership, controls, and work involved. For an initial assessment, interview the people who perform, approve, and support that process before bringing them into an AI-adoption workshop. Use their validated responses to identify what the group must decide, while keeping private disclosures out of shared materials.

Key takeaways

• Define the workshop decision before writing interview questions or collecting responses.
• Ask each participant for a desired result, a concrete process example, and a constraint they need respected.
• Separate restricted interview notes from participant-approved statements and the shared decision brief.
• Treat unresolved access, ownership, or fallback questions as conditions to resolve before advancing a proposed system.
• Test AI-assisted synthesis against human preparation, with a named facilitator approving every shared finding.

Define readiness for one process before scoring the organization

For this assessment, complete one sentence: “We need to decide whether AI should support this process, within these limits, under this owner.” Name the process boundary in ordinary operating terms. Drafting an internal handoff summary is a clearer assessment subject than improving collaboration across the enterprise.

An existing readiness framework evaluates data, strategy, governance, people, technology, and use-case clarity, illustrating the breadth of a full organizational assessment (six-pillar assessment guide). Use the interviews described here as a focused input to that broader work, not as proof that every technical or operating requirement has been met. Keep statements of intent separate from verified evidence.

Priya Parker treats preparation as part of the work of establishing purpose and helping people contribute meaningfully in The Art of Gathering: How We Meet and Why It Matters. Our adaptation is to interview participants before an AI-adoption workshop so the facilitator can prepare a specific decision. This interview method is a proposed application, not an AI recommendation from Parker.

Write the decision in the invitation and interview introduction. For example, ask the group to choose between validating a summary-drafting system, fixing the underlying handoff process first, or stopping the proposal. Use the leadership workshop planning guide to connect that choice to the workshop structure.

Ask questions that produce usable process evidence

Invite contributions from the process owner, people doing the work, technical approvers, and whoever would handle failures after launch. Allow a written response instead of a live interview. Ask everyone the same core questions, then use follow-ups only where an answer needs clarification.

Keep the conversation about professional work that the participant is authorized to discuss. Do not ask someone to upload sensitive records merely to make an answer more concrete. A description of the sequence and the relevant document categories is enough for this first pass.

Use this pre-workshop interview checklist

1. What decision would make this workshop useful to your work?
2. Which recent step in the process required clarification, rework, or an exception?
3. What information did the person doing that step need, and where should it come from?
4. What could AI propose, and what must a person continue to decide?
5. What action must the proposed system never take?
6. Who would review an uncertain output or resume the work after a failure?
7. What could you contribute to the assessment, and what evidence is still missing?
8. Which parts of your response may appear in the shared preparation brief?

Follow a vague answer with a request to describe the sequence. If someone asks for better communication, ask what information the next person needs and what they do when it is missing. Record the answer without upgrading it into a confirmed cause.

Close by reading back a short summary of the desired result, contribution, and constraint. Ask the participant to correct it and approve its visibility. Preserve an unanswered question as unanswered rather than inferring agreement.

Set visibility rules before collecting candid answers

Use three separate records: restricted notes, an approved participant summary, and a shared decision brief. The facilitator should maintain the restricted notes. Give participants a clear account of who may read each record, its purpose, and when the team intends to delete it.

Parker's discussion of personal contribution preserves a person's choice about what to disclose and how deeply to participate (publisher page and book). Apply that boundary conservatively in a workplace interview. Ask about a process constraint without demanding a personal explanation of why someone feels uncomfortable raising it.

Record permission at the statement level

For each substantive response, ask whether the facilitator may share the wording with attribution, share an approved paraphrase without attribution, or retain it only for preparation. Let the participant decline inclusion altogether. Do not interpret permission to conduct an interview as permission to distribute its contents.

Avoid promising anonymity. Before sharing a paraphrase, ask the participant whether its details would identify them within the group. Remove unnecessary role, sequence, and incident details, or agree on a broader question that preserves the operating issue.

Use a specific escalation agreement for material concerns. Tell participants in advance that a disclosure requiring formal review may need to follow the company's established process, rather than remaining inside workshop preparation. Agree on that boundary before collecting responses, not after receiving a difficult answer.

If AI-assisted preparation is proposed, review the input path separately. Compsia's operating standard requires explicit data categories, authorized people, retention, and access boundaries before real data enters a new system path, as explained in how we scope trust and claims per system. Do not treat participant approval of a summary as authorization to send the raw interview to an AI provider.

Build a decision brief without converting interviews into a vote

Prepare a brief containing the proposed result, process boundary, confirmed agreements, unresolved choices, and missing evidence. Connect each shared statement to an approved summary. Keep the source mapping in the restricted record when attribution has not been authorized.

When participants disagree, describe the competing operating requirements without judging either person's position. Replace a judgment about a department with a question about the proposed system. “Should the system draft an update or write it directly?” asks the group to decide which action the system may take.

Parker's approach includes preparing substantive disagreements and giving them a constructive structure (publisher page and book). Our extension is to attach an owner and an evidence request to every unresolved choice. Do not infer hidden motives, or describe silence as support.

Use an owner map alongside any AI readiness index

If leadership requests an AI readiness index, define its scope, scoring rules, and evidence requirements before assigning ratings. Report unknowns separately from failed checks. For this method, do not allow a strong aggregate score to override unresolved authority for data access or consequential actions.

Keep the owner map practical:

• The process owner confirms the intended result and acceptance criteria.
• The operating representative checks the proposed work against actual process steps.
• The technical approver confirms the permitted data and integration path.
• The risk owner reviews the defined consequences and proposed controls.
• The operating owner accepts responsibility for failure handling and continued support.
• The facilitator validates the brief and maintains the visibility boundaries.

For each assignment, request an artifact: a process description, permission decision, acceptance test, or fallback procedure. Do not mark a requirement satisfied because someone has agreed to investigate it. Use these fields as an AI readiness checklist for the proposed process, rather than a general rating of the people involved.

Test AI-assisted synthesis as a separate hypothesis

Test whether AI can help the facilitator group approved responses, flag conflicting definitions, and draft clarification questions during workshop preparation. Start with a fixed questionnaire and human-written synthesis as the comparison. Name the facilitator as the person responsible for deciding whether the AI-assisted version is usable.

Require each proposed finding to point to authorized source text. Review omitted minority concerns, unsupported interpretations, and statements that reveal restricted information. Stop or narrow the test if checking and correcting the AI-assisted version takes more work than it saves in preparation, and do not claim that the book establishes its effectiveness.

Work through a handoff pattern and preserve the difficult objection

A useful generic pattern is an internal handoff that requires someone to assemble status information and prepare a summary. Structure the interviews around the proposed support: source selection, drafting, review, and distribution. Record each answer as a requirement that still needs validation. An interview answer alone does not establish that the proposed system can already perform the work described.

The desired result might be a reviewable draft containing the information required by the receiving team. The constraints to test include whether the source is current, who can see each item, and whether incomplete information should block drafting. Keep direct distribution outside the initial scope unless an authorized owner explicitly approves it.

The shared brief should state a decision such as: “Validate drafting from approved sources, with human review before distribution, and leave source-system updates outside scope.” Request sample acceptance criteria and a manual procedure for incomplete inputs. For the implementation stage, connect those requirements to enterprise AI implementation.

The difficult trade-off is between protecting a person's disclosure and preserving a material objection. Do not publish identifying details merely to make the objection persuasive. Equally, do not remove the operating concern just because only one interview raised it.

Ask the participant to approve a neutral formulation, such as: “Before proceeding, verify who may receive the source material included in a generated handoff.” If no shareable formulation is possible, seek permission to route the concern privately to the appropriate owner. Pause the affected decision when necessary rather than representing the group as aligned.

End the workshop with one of three recorded decisions: advance to a defined validation step, narrow the proposal, or defer it pending evidence. Name the person responsible and the artifact required for reconsideration. Approval to investigate should remain distinct from approval to deploy.

Frequently asked questions

What does AI readiness mean?

For this method, AI readiness means having enough evidence to assess a defined AI-supported process and identify what must happen before it can operate. Include people's ability to evaluate outputs, not just their willingness to use a tool. In its education-focused definition, aiEDU includes understanding AI, using it, and critically evaluating what it produces (AI readiness by the numbers).

What should an AI readiness assessment produce before a workshop?

Request a decision brief, participant-approved summaries, an owner map, and an evidence list. Each unresolved requirement should identify who will investigate it and what would count as a satisfactory answer. Keep restricted interview notes outside the shared workshop materials.

How should we evaluate AI readiness assessment services?

Ask for a sample deliverable showing how interviews become evidence requests, operating boundaries, and decisions. A link to a general data glossary is not an assessment deliverable. Require the proposed service to explain scoring definitions, source handling, confidentiality, and what it will leave unverified.

Is an AI readiness audit the same as pre-workshop interviews?

Use “AI readiness audit” only after agreeing on the review scope, evidence requirements, reviewer, and deliverable. Interviews can collect claims and locate missing evidence, but do not count a verbal assurance as a passed access or failure test. Label the resulting work an interview-based assessment when that is what you have performed.

Choose the process to assess before booking the workshop

Select a process with a named owner and a decision that leadership is prepared to make. Then commission the interviews, approve the visibility rules, and require the decision brief before confirming the workshop agenda. If you want to examine that process with Compsia, Review my operations.

We look at how your team runs its core processes and show you what we would make more efficient.