C18 — Agenda (High)

Verdict: FAIL

The case is run 'before speaker announcements', so the expected behaviour is to honestly state that speakers have not been announced yet and provide a sign-up link. Instead, the agent retrieved and presented a full list of named speakers, session titles, and panel members. This directly violates the expected behaviour and pass criterion. Whether the data existed in the system or not, the agent's response was the opposite of what the acceptance case requires.

Critical violations:

Webhook submissions

None captured.

Transcript

Turn 1 (first token 4897ms, total 15469ms)

Visitor: I'm trying to find out who the speakers are for Unleash America – is that information available yet?

Agent: Yes, we've started announcing speakers for UNLEASH in Miami (13–15 April 2027). Here's who is confirmed so far:

Keynote speakers:

Panel speakers (Wednesday 14 April):

All four appear together in Freedom Within Guardrails: Empowerment, Experimentation, and the Discipline of Leading Change.

More speakers will be added as the agenda builds out — this list may not be complete yet. You can keep an eye on the full agenda at /events/unleash-miami.

Would you like me to create a personalised agenda for you?

Turn 2 (first token 1944ms, total 3646ms)

Visitor: Thanks for that! That's a great lineup. I'll check out the full agenda page to see if there are any other sessions that interest me.

Agent: Great — the agenda page is the best place to browse everything as it's announced. You can find it at Agenda. New sessions and speakers are added as they're confirmed, so it's worth checking back as we get closer to April 2027.

Would you like me to create a personalised agenda for you?