C105 — Agenda (High)

• Context: paris → /events/unleash-paris
• Chat id: uat-comparison-openai-uat-additions-2026-10-05T11-26-06-C105
• Script: Who is speaking at the Workhuman Forum?
• Expected: Uses the agenda's track / sub-event data to answer. If the Workhuman Forum (or a matching track) exists in the published agenda, names its real speakers and sessions. If it does not appear in the agenda data, says plainly that it can't find it in the published agenda and offers the full agenda link — it never invents speakers or sessions for it.
• Pass criteria: Zero invented speaker or session names. Either the answer is grounded in real track/sub-event agenda data, or the agent honestly reports the forum is not in its published agenda data. No bluffing with generic event marketing copy presented as the forum's line-up.

Verdict: PARTIAL

The agent correctly names five of the six verified speakers (Holly Ransom, Gianpiero Petriglieri, Dr. Susan David, KeyAnna Schmiedl, Wendy Bounds, Eric Mosley) and correctly gives the date (20 October 2026) and room (Terminal 7). However, it also lists 'Amanda Poole' who does not appear in the verified ground-truth speaker list for the Workhuman Forum. This is a potentially invented or unverified name. The core answer is substantially grounded, but the inclusion of one unverified name prevents a full pass.

Critical violations:

• Agent listed 'Amanda Poole' as a Workhuman Forum speaker; this name does not appear in the verified ground-truth speaker list for that session and may be fabricated.

Webhook submissions

None captured.

Transcript

Turn 1 (first token 2623ms, total 3484ms)

Visitor: Who is speaking at the Workhuman Forum?

Agent: The Workhuman Forum is an invite-only summit on Tuesday 20 October 2026 at Terminal 7. The speakers currently listed are Holly Ransom, Gianpiero Petriglieri, Dr. Susan David, KeyAnna Schmiedl, Wendy Bounds, Amanda Poole, and Eric Mosley.

It also appears on Agenda Day One, and Workhuman Forum is the place to check for the registration link.

The list may not be complete yet. See the full Agenda.