---
schema: 1
type: video
title: 'Why building eval platforms is hard — Phil Hetzel, Braintrust'
description: >-
  An eval platform is not just a test runner. You are building shared
  definitions of "good," reliable data pipelines, labelling workflows,
  versioning, and trust in results across many teams and model changes. This
  session breaks down the hidden complexity, the common failure modes, and the
  design principles that make evals credible and usable in day-to-day
  engineering. Speaker info: - https://www.linkedin.com/in/philliphetzel/ ---
  Watch along with the corrected transcript, summary, timestamps and resources
  on this talk's official AIE page:
  https://ai.engineer/talks/_fQ7Z_Wfouk-why-building-eval-platforms-is-hard 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/channel/aidotengineer/explore/video/why-building-eval-platforms-is-hard-phil-hetzel-bra-fc748e3d63d7
markdown: >-
  https://liquidgold.cc/channel/aidotengineer/video/why-building-eval-platforms-is-hard-phil-hetzel-bra-fc748e3d63d7.md
llms: null
parent: 'https://liquidgold.cc/channel/aidotengineer.md'
channel: aidotengineer
updated: '2026-10-24T14:43:24.829Z'
language: null
children: []
videoId: _fQ7Z_Wfouk
durationSeconds: 1539
publishedAt: '2026-04-28T16:00:06.000Z'
searchChunks:
  - ordinal: 0
    text: null
    startSeconds: 125
    endSeconds: 235
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=125'
    heldReason: rights
  - ordinal: 1
    text: null
    startSeconds: 245
    endSeconds: 346
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=245'
    heldReason: rights
  - ordinal: 2
    text: null
    startSeconds: 458
    endSeconds: 582
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=458'
    heldReason: rights
  - ordinal: 3
    text: null
    startSeconds: 592
    endSeconds: 731
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=592'
    heldReason: rights
  - ordinal: 4
    text: null
    startSeconds: 742
    endSeconds: 823
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=742'
    heldReason: rights
  - ordinal: 5
    text: null
    startSeconds: 833
    endSeconds: 964
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=833'
    heldReason: rights
  - ordinal: 6
    text: null
    startSeconds: 833
    endSeconds: 964
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=833'
    heldReason: rights
  - ordinal: 7
    text: null
    startSeconds: 974
    endSeconds: 1063
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=974'
    heldReason: rights
  - ordinal: 8
    text: null
    startSeconds: 1215
    endSeconds: 1306
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1215'
    heldReason: rights
  - ordinal: 9
    text: null
    startSeconds: 1063
    endSeconds: 1074
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1063'
    heldReason: rights
  - ordinal: 10
    text: null
    startSeconds: 1074
    endSeconds: 1205
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1074'
    heldReason: rights
  - ordinal: 11
    text: null
    startSeconds: 1316
    endSeconds: 1415
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1316'
    heldReason: rights
  - ordinal: 12
    text: null
    startSeconds: 1316
    endSeconds: 1415
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1316'
    heldReason: rights
  - ordinal: 13
    text: null
    startSeconds: 1415
    endSeconds: 1425
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1415'
    heldReason: rights
  - ordinal: 14
    text: null
    startSeconds: 1425
    endSeconds: 1539
    watch: 'https://liquidgold.cc/watch/_fQ7Z_Wfouk?t=1425'
    heldReason: rights
---


# Why building eval platforms is hard — Phil Hetzel, Braintrust

An eval platform is not just a test runner. You are building shared definitions of "good," reliable data pipelines, labelling workflows, versioning, and trust in results across many teams and model changes. This session breaks down the hidden complexity, the common failure modes, and the design principles that make evals credible and usable in day-to-day engineering. Speaker info: - https://www.linkedin.com/in/philliphetzel/ --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/\_fQ7Z_Wfouk-why-building-eval-platforms-is-hard 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/channel/aidotengineer/explore/video/why-building-eval-platforms-is-hard-phil-hetzel-bra-fc748e3d63d7 · 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.