An AI consulting discovery call is a diagnostic conversation, not a product demo. Your job is to understand the business problem, current workflow, cost or consequence, data and systems, risk, stakeholders, decision process and desired outcome well enough to decide whether there is a sensible next step.

Do not solve the whole project on the call. Give enough insight to demonstrate judgment, but discovery should determine whether an engagement makes sense—not become a free implementation workshop.

Before the call: do 10–15 minutes of homework

Review the company's website, business model, public technology signals, recent changes and the person's role. Form one or two hypotheses about where the problem might live, but hold them loosely. Good preparation helps you ask sharper questions; it should not cause you to force the prospect into a preselected solution.

Open by setting the frame

A simple opening is: “I'd like to understand what's happening today, what you're trying to improve and what you've already tried. If it looks like there is a sensible next step, we can talk about that at the end.”

This gives the conversation structure and removes pressure to immediately pitch tools.

15 AI consulting discovery questions

  1. What triggered this conversation now? Look for a real change, deadline, cost or executive priority.
  2. What business outcome are you trying to improve? Keep the conversation anchored in the outcome rather than “using AI.”
  3. Walk me through the current workflow from start to finish.
  4. Who touches the process, and where does work slow down or get repeated?
  5. How often does this happen and at what volume?
  6. What does the problem cost today—in time, money, quality, risk or customer experience?
  7. What have you already tried? This reveals previous failures and internal constraints.
  8. Which systems are involved? Ask where information enters, moves and ends up.
  9. What data would a solution need, and who owns access to it?
  10. What happens if the AI is wrong? The consequence determines the level of human review and testing.
  11. Which decisions must remain human-owned?
  12. Who would use the new workflow day to day? Adoption can matter as much as technical feasibility.
  13. Who else needs to approve the project?
  14. Is there a budget range or approval process we should design around?
  15. What would have to be true for this to feel successful 60 or 90 days from now?

Specialist discovery resources emphasize the same pattern: begin with operational pain, then qualify readiness, data, budget and decision structure rather than jumping straight into a solution. Auditic's discovery-question guide and ConsultKit's AI discovery framework are useful independent examples.

What to listen for in the answers

SignalWhat it may mean
Named workflow + measurable painThere may be a real project to assess.
“We just need AI somewhere”Start with opportunity assessment, not implementation.
No access to required dataReadiness work may come before a pilot.
High consequence if wrongIncrease human review, testing and governance—or avoid automation.
No owner or sponsorThe project may stall regardless of technology.
Clear outcome + sponsor + bounded workflowA scoped assessment or pilot may be appropriate.

Our AI readiness assessment can help when the conversation reveals that the business has interest but not yet the conditions for implementation.

Diagnose before you prescribe

A consultant earns trust by being willing to conclude that AI is not the first answer. Sometimes the problem is poor process design, missing data, unclear ownership or a conventional automation opportunity. Use the AI opportunity assessment framework to separate attractive demos from useful business cases.

How much advice should you give away?

You should demonstrate that you understand the problem and can think clearly about it. You do not need to design the entire architecture, map every prompt, select every vendor and hand over a complete implementation plan for free.

A useful boundary is to discuss what needs to be understood and why, while reserving detailed solution design for a paid assessment or project. If a prospect needs a full roadmap before committing to implementation, the roadmap itself can be the engagement.

Want a more structured path? The AI Consultant Certification program we refer combines AI training with consulting-business topics including client acquisition, discovery calls, proposals and setting up a consulting practice. Structured training is optional—you can build these skills independently—but it may suit someone who wants a guided curriculum and accountability. Affiliate disclosure: we may earn a commission on an eligible purchase, at no additional cost to you. Training does not guarantee clients, income or business results.

How to close the discovery call

Summarize what you heard: the problem, impact, constraints and desired outcome. Then choose one of three endings:

If a proposal is appropriate, use our AI consulting proposal template to turn the discovery inputs into scope, deliverables, milestones, success measures and price.

A simple note-taking structure

Use seven headings: Trigger / Problem / Workflow / Impact / Data & Systems / Risk & People / Decision & Next Step. This is enough structure to keep the call useful without making it feel like an interrogation.

The bottom line

A strong AI consulting discovery call should leave both sides clearer—not merely more excited about AI. Understand the workflow, quantify the problem, test readiness, surface risk and decision dynamics, and only then decide whether a proposal belongs on the table.