---
title: "Your Agent Is Lying to You About Whether It Worked"
category: "talks"
date: "2026-06-29"
time: "12:05pm-12:25pm"
track: "Expo Stage 1 NE"
room: "Expo Stage 1 NE"
speakers: ["Dat Ngo"]
sourceLabels: ["Official conference schedule", "Public YouTube metadata"]
scheduleTrack: ""
scheduleRoom: "Expo Stage 1 NE"
scheduleLabels: ["Expo Stage 1 NE", "session", "confirmed"]
---
# Your Agent Is Lying to You About Whether It Worked

## Conference Context
- Date/time: 2026-06-29 · 12:05pm-12:25pm
- Track/room: track TBD · Expo Stage 1 NE
- Speaker(s): Dat Ngo
- Session type/status: session · confirmed

- Track: track TBD
- Room: Expo Stage 1 NE
- Session type: session
- Status: confirmed

## Session Description
Every span is green, every tool call returned cleanly, and the agent still regenerated the same plan 27 times before giving up invisible to any outcome metric, obvious in the trajectory. We pull up a real trace where the outcome looks healthy and the path is a disaster, then show Signal, our agent, surfacing it automatically: sweeping the project, ranking it above the noise, and linking straight to the offending trace with debugging evidence attached. The live version of the trajectory-over-outcomes argument, with a one-click path from "something's wrong" to "here's exactly where."

## Media Evidence
[LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize](https://www.youtube.com/watch?v=JsCCrBF7F1g) (speaker-match related prior/adjacent AI Engineer video; captions: English auto-captions).

- Source video: `youtube-JsCCrBF7F1g`
- Slide deck: [[youtube-JsCCrBF7F1g-dense-slides|Dense Slides: LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize]] — 1 visible slide image(s); 1 HTML recreation(s).
![[assets/dense-slides/JsCCrBF7F1g/slide-001.jpg]]
- Additional slide evidence: [[youtube-JsCCrBF7F1g-slides|Slides: LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize]], [[youtube-JsCCrBF7F1g-reconstructed-slides|Reconstructed Slides: LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize]]
- Slide-derived themes for `youtube-JsCCrBF7F1g`: happening, application, root, cause, down, problem, well, product.

## Evidence Graph
This evidence graph is generated from currently linked source material: official schedule text, related video pages, cached transcripts, visible slide text, dense/reconstructed slide pages, and AI slide-classification audits.

### Media Signals
- `youtube-JsCCrBF7F1g` — 4 slide-derived text signals
- Slide-derived themes for `youtube-JsCCrBF7F1g`: happening, application, root, cause, down, problem, well, product.
- Evidence links for `youtube-JsCCrBF7F1g`: [[youtube-JsCCrBF7F1g]], [[youtube-JsCCrBF7F1g-slides]], [[youtube-JsCCrBF7F1g-dense-slides]], [[youtube-JsCCrBF7F1g-reconstructed-slides]]

### Agent Reading Notes
Use these signals to refine the synopsis, topic links, people/company context, and method notes. If a source is a related external video rather than an exact official recording, keep it framed as supporting evidence.

## Transcript Status
Related video transcript availability: English auto-captions. Treat this as supporting context, not a recording of this exact scheduled session unless later confirmed. Not fetched yet.

## People
- [[dat-ngo]]

## Supporting Slides
- [[youtube-JsCCrBF7F1g-slides]] — extracted from the related public AI Engineer video.

## Slide Evidence
- Slide-only cropped deck: [[youtube-JsCCrBF7F1g-dense-slides]] (1 viable slide images).
- Related slide/OCR pages:
- [[youtube-JsCCrBF7F1g-dense-slides]]
- [[youtube-JsCCrBF7F1g-reconstructed-slides]]
- [[youtube-JsCCrBF7F1g-slides]]
- Slide-derived terms: `observability`, `evaluation`, `teams`, `engineer`, `europe`, `braintrust`, `workos`, `openal`, `experimentation`, `genal`, `hard`, `tackling`, `know`, `sevan`, `arize`, `platform`, `google`, `deepmind`

## Synthesis
### Synthesized Breakdown
# Your Agent Is Lying to You About Whether It Worked ## Conference Context - Date/time: 2026-06-29 · 12:05pm-12:25pm - Track/room: track TBD · Expo Stage 1 NE - Speaker(s): Dat Ngo - Session type/status: session · confirmed - Track: track TBD - Room: Expo Stage 1 NE - Session type: session - Status: confirmed ## Session Description Every span is green, every tool call returned cleanly, and the agent still regenerated the same plan 27 times before giving up invisible to any outcome metric, obvious in the trajectory. We pull up a real trace where the outcome looks healthy and the path is a disaster, then show Signal, our agent, surfacing it automatically: sweeping the project, ranking it above the noise, and linking straight to the offending trace with debugging evidence attached. The live version of the trajectory-over-outcomes argument, with a one-click path from "something's wrong" to "here's exactly where." ## Media Evidence [LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize](https://www.youtube.com/watch?v=JsCCrBF7F1g) (speaker-match related prior/adjacent AI Engineer video; captions: English auto-captions). - Source video: `youtube-JsCCrBF7F1g` - Slide deck: [[youtube-JsCCrBF7F1g-dense-slides|Dense Slides: LLM Observability, Evaluation, Experimentation Platform — Dat Ngo, Arize]] — 1 visible slide image(s); 1 HTML recreation(s).

### Speaker And Company Context
- [[dat-ngo|Dat Ngo]] — AI Architect at [[arize-ai|Arize AI]].

### Topics Covered
- [[agent-security]]

### Derived Links And Source Material
- [[youtube-JsCCrBF7F1g]] — related YouTube source page.
- [[youtube-JsCCrBF7F1g-slides]] — slide evidence.
- [[youtube-JsCCrBF7F1g-reconstructed-slides]] — slide evidence.
- [[youtube-JsCCrBF7F1g-dense-slides]] — slide evidence.

### Novel Concepts / Clever Methods
- No highlighted novel concept has been detected yet.

### Evidence Boundary
This synthesis is based on the official schedule and linked source pages. It should be revisited when exact session recordings or transcript-backed secondary sources are available.