A10XAI FDE in 2 weeks

Free · a real FDE interview round, scored instantly

Can you pass a real FDE interview case? Try it, it's below.

The FDE Rapid Round · 1/7

A customer's support agent demos perfectly. In production it answers wrong about 1 time in 8. You have one week and read access to their traffic logs.

What is your first move?

Rather write the full case? Start it below ↓

Free · no signup to try60 min · sierra-style loop

Meridian Health: L1 agent that must not invent clinical advice

The situation, in 20 seconds

  • Who: Priya, VP CX (customer experience) at Meridian Health — 14 hospitals, ~38k chat/email tickets a month on Zendesk, plus a voice IVR (phone menu) still dumping to a 90-person BPO (outsourced call center).
  • What she wants: an AI agent live on web chat in 10 weeks that "deflects 40% of L1" (first-line support tickets) so she can cut BPO spend before the next board meeting. Marketing already announced it on the patient portal.
  • Her metric: containment rate, the share of conversations the agent resolves without a human. She has no labeled intent taxonomy, just 40 sticky notes from a workshop.
  • Hard constraints: Legal says anything clinical must escalate. IT allows read-only FHIR (the standard healthcare-records API) for demographics and appointments in v1. The Zendesk sandbox has 3 years of tickets with PHI (protected health information) in free text.
  • Politics and budget: front-desk staff are circulating a petition that AI will "give medical advice." The CMO (chief medical officer) is nervous and not in the room. Licenses approved, plus one internal engineer at 50%.
  • She asks: how would you scope the pilot, what would you build, and when can we go live?
Read the full scenario

You're on a call with Priya, VP CX (customer experience) at Meridian Health, a 14-hospital regional system. They run ~38k chat/email tickets per month on Zendesk, plus a separate voice IVR (phone menu) that still dumps to a 90-person BPO (outsourced call center). First-response SLA (service-level agreement) is 15 minutes for chat; the CSAT (customer satisfaction) target is 4.4/5. Average handle time is 11 minutes. Top intents they think they know: appointment reschedule (~22%), billing explanation (~18%), portal password resets (~12%), referral status (~9%), and a long tail of clinical-adjacent questions ('is this medication covered,' 'what does this lab mean'). Priya wants an AI agent live on web chat in 10 weeks that 'deflects 40% of L1' (first-line support tickets) so she can cut BPO spend before the next board meeting. Marketing already announced 'AI Care Concierge' on the patient portal homepage. The CMO (chief medical officer) is nervous but not in the room. Ops claims macros and FAQs are 'mostly clean'; Legal says anything clinical must escalate. IT will only allow read-only FHIR (the standard healthcare-records API) access to Epic for demographics/appointments in v1, and the Zendesk sandbox has three years of tickets with PHI (protected health information) still in free-text notes. Her success metric is 'containment rate,' the share of conversations the agent resolves without a human. She does not have a labeled intent taxonomy, only 40 sticky-note categories from a workshop. Unionized hospital front-desk staff are already circulating a petition that AI will 'give medical advice.' Budget is approved for licenses and one internal engineer at 50% for the pilot. She asks: how would you scope the pilot, what would you build, and when can we go live?

You're on a call with Priya, VP CX (customer experience) at Meridian Health, a 14-hospital regional system. They run ~38k chat/email tickets per month on Zendesk, plus a separate voice IVR (phone menu) that still dumps to a 90-person BPO (outsourced call center). First-response SLA (service-level agreement) is 15 minutes for chat; the CSAT (customer satisfaction) target is 4.4/5. Average handle time is 11 minutes. Top intents they think they know: appointment reschedule (~22%), billing explanation (~18%), portal password resets (~12%), referral status (~9%), and a long tail of clinical-adjacent questions ('is this medication covered,' 'what does this lab mean'). Priya wants an AI agent live on web chat in 10 weeks that 'deflects 40% of L1' (first-line support tickets) so she can cut BPO spend before the next board meeting. Marketing already announced 'AI Care Concierge' on the patient portal homepage. The CMO (chief medical officer) is nervous but not in the room. Ops claims macros and FAQs are 'mostly clean'; Legal says anything clinical must escalate. IT will only allow read-only FHIR (the standard healthcare-records API) access to Epic for demographics/appointments in v1, and the Zendesk sandbox has three years of tickets with PHI (protected health information) still in free-text notes. Her success metric is 'containment rate,' the share of conversations the agent resolves without a human. She does not have a labeled intent taxonomy, only 40 sticky-note categories from a workshop. Unionized hospital front-desk staff are already circulating a petition that AI will 'give medical advice.' Budget is approved for licenses and one internal engineer at 50% for the pilot. She asks: how would you scope the pilot, what would you build, and when can we go live?

This is the round that fails the most FDE candidates. Answer in the boxes below. The clock starts when you start typing. Then get graded free against the hidden rubric top labs use.

Meridian Health: L1 agent that must not invent clinical advice60:00Clock starts when you type. No signup.

Scope & clarifying questions

Architecture

Eval plan

Rollout & risks

Asks

Your answer

The rubric grades all five. Keep them in view as you write:
1 Scope & clarifying questions · 2 Architecture · 3 Eval plan · 4 Rollout & risks · 5 Asks

Same case, same rubric, same grader as the desk version. On a laptop you get five separate boxes.

Not right now? Save the case.

We'll email you the link so you can run it whenever you have a clear hour, plus the Monday brief that keeps you current on the FDE field.