---
schema: 1
type: video
title: >-
  Mind the Gap (In your Agent Observability) — Amy Boyd & Nitya Narasimhan,
  Microsoft
description: >-
  Agents drift. Models change, prompts get tweaked, edge cases accumulate, and
  the gap between what your agent does and what you need it to do widens without
  you noticing. Amy and Nitya walk through Microsoft Foundry's observability
  stack: tracing built on OpenTelemetry, built-in evaluators for quality,
  safety, and agentic metrics like intent resolution and task adherence, and red
  teaming where a second AI attacks your agent with adversarial prompts to find
  vulnerabilities before your users do. The piece worth watching for is the
  observe skill demo. You point it at an agent with no eval dataset, no
  baselines, nothing. It generates the dataset, runs batch evaluations,
  optimizes the prompt, compares versions, and rolls back to the best one... all
  from a single prompt to a coding agent. The skill shows its reasoning at each
  step, which is where the real value is: it surfaces the failures you didn't
  know to look for. Speaker info: - https://x.com/NityaNarasimhan -
  https://www.linkedin.com/in/nityan/ - https://x.com/AmyKateNicho -
  https://www.linkedin.com/in/amykatenicho/ --- Watch along with the corrected
  transcript, summary, timestamps and resources on this talk's official AIE
  page:
  https://ai.engineer/talks/iOXM3zE-2dk-mind-the-gap-in-your-agent-observability
  Join our upcoming AI Engineer conferences (see https://ai.engineer landing
  page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs
  and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference
  returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS
  2026 in Sydney ---
canonical: 'https://liquidgold.cc/watch/iOXM3zE-2dk'
markdown: 'https://liquidgold.cc/watch/iOXM3zE-2dk.md'
llms: 'https://liquidgold.cc/watch/iOXM3zE-2dk/llms.txt'
parent: 'https://liquidgold.cc/channel/aidotengineer.md'
channel: aidotengineer
updated: '2026-10-24T14:11:12.808Z'
language: null
children: []
videoId: iOXM3zE-2dk
durationSeconds: 4807
publishedAt: '2026-05-14T16:00:06.000Z'
searchChunks:
  - ordinal: 0
    text: null
    startSeconds: 114
    endSeconds: 234
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=114'
    heldReason: rights
  - ordinal: 1
    text: null
    startSeconds: 246
    endSeconds: 355
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=246'
    heldReason: rights
  - ordinal: 2
    text: null
    startSeconds: 365
    endSeconds: 469
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=365'
    heldReason: rights
  - ordinal: 3
    text: null
    startSeconds: 480
    endSeconds: 592
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=480'
    heldReason: rights
  - ordinal: 4
    text: null
    startSeconds: 731
    endSeconds: 741
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=731'
    heldReason: rights
  - ordinal: 5
    text: null
    startSeconds: 885
    endSeconds: 983
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=885'
    heldReason: rights
  - ordinal: 6
    text: null
    startSeconds: 994
    endSeconds: 1104
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=994'
    heldReason: rights
  - ordinal: 7
    text: null
    startSeconds: 1338
    endSeconds: 1349
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=1338'
    heldReason: rights
  - ordinal: 8
    text: null
    startSeconds: 1807
    endSeconds: 1819
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=1807'
    heldReason: rights
  - ordinal: 9
    text: null
    startSeconds: 2039
    endSeconds: 2136
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=2039'
    heldReason: rights
  - ordinal: 10
    text: null
    startSeconds: 2385
    endSeconds: 2395
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=2385'
    heldReason: rights
  - ordinal: 11
    text: null
    startSeconds: 2797
    endSeconds: 2919
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=2797'
    heldReason: rights
  - ordinal: 12
    text: null
    startSeconds: 2930
    endSeconds: 3051
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=2930'
    heldReason: rights
  - ordinal: 13
    text: null
    startSeconds: 3061
    endSeconds: 3179
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3061'
    heldReason: rights
  - ordinal: 14
    text: null
    startSeconds: 3300
    endSeconds: 3312
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3300'
    heldReason: rights
  - ordinal: 15
    text: null
    startSeconds: 3535
    endSeconds: 3546
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3535'
    heldReason: rights
  - ordinal: 16
    text: null
    startSeconds: 3645
    endSeconds: 3756
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3645'
    heldReason: rights
  - ordinal: 17
    text: null
    startSeconds: 3767
    endSeconds: 3856
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3767'
    heldReason: rights
  - ordinal: 18
    text: null
    startSeconds: 3948
    endSeconds: 3958
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=3948'
    heldReason: rights
  - ordinal: 19
    text: null
    startSeconds: 4058
    endSeconds: 4158
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=4058'
    heldReason: rights
  - ordinal: 20
    text: null
    startSeconds: 4501
    endSeconds: 4621
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=4501'
    heldReason: rights
  - ordinal: 21
    text: null
    startSeconds: 4632
    endSeconds: 4744
    watch: 'https://liquidgold.cc/watch/iOXM3zE-2dk?t=4632'
    heldReason: rights
---


# Mind the Gap (In your Agent Observability) — Amy Boyd & Nitya Narasimhan, Microsoft

Agents drift. Models change, prompts get tweaked, edge cases accumulate, and the gap between what your agent does and what you need it to do widens without you noticing. Amy and Nitya walk through Microsoft Foundry's observability stack: tracing built on OpenTelemetry, built-in evaluators for quality, safety, and agentic metrics like intent resolution and task adherence, and red teaming where a second AI attacks your agent with adversarial prompts to find vulnerabilities before your users do. The piece worth watching for is the observe skill demo. You point it at an agent with no eval dataset, no baselines, nothing. It generates the dataset, runs batch evaluations, optimizes the prompt, compares versions, and rolls back to the best one... all from a single prompt to a coding agent. The skill shows its reasoning at each step, which is where the real value is: it surfaces the failures you didn't know to look for. Speaker info: - https://x.com/NityaNarasimhan - https://www.linkedin.com/in/nityan/ - https://x.com/AmyKateNicho - https://www.linkedin.com/in/amykatenicho/ --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/iOXM3zE-2dk-mind-the-gap-in-your-agent-observability Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney ---

Canonical: https://liquidgold.cc/watch/iOXM3zE-2dk · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md)

> Search chunk text is held until public-source provenance is verified.

## Document chunks

> Missing: no stored document chunks are available.
