TechPrep
Mock roundsReadLabPracticeCheck

Mock: customer case (FDE and Applied AI)

45 minutes: a partner plays a customer with a vague AI goal; you run discovery, scope a solution, design it and handle an objection.

This mirrors what FDE and Applied AI posts describe: discovery, architecture, evaluation and stakeholder communication (Sources A6, A7, O6, O7). The case below is written for this site, not a reported interview.

Partner brief. You are the VP of Operations at an insurer. "We want AI to speed up claims." You know: 40,000 claims a month, adjusters spend ~30% of time reading documents, data is in a claims system plus a document store, legal is nervous, and IT wants it in your Azure tenancy. Reveal facts only when asked. At minute 35, object: "How do I know it won't make a wrong decision on a claim?"

This rubric is ours, built from published signals (for example OpenAI's "well-designed solutions, high-quality code, optimal performance, and good test coverage"). It is not any company's scoresheet. See Sources.

Customer case

45:00
  1. Discovery (15 min)
  2. Scope and success criteria (8 min)
  3. Solution design (12 min)
  4. Objection and close (10 min)

Score yourself (or have a partner score you)

Discovery Found the real problem and metric
Scoping A shippable first use case
Technical design Sound, appropriate
Objection handling Trust and risk
Communication Executive-ready

Past attempts

Practice

Written for this site to drill the lesson. Not reported interview questions.

  1. Swap roles and play the customer; note what a strong candidate asked that you did not.

Check yourself

What should the first use case come with?

Success criteria (metric, baseline, threshold) and a path to production.

Saved in this browser.