---
schema: 1
type: video
title: Agents Need Feature Flags - Sachin Gupta
description: >-
  Most AI teams ship behavior changes to 100% of users on every deploy — no
  canary, no segment, no kill switch. Web teams stopped doing this in 2012. AI
  teams are about to learn why. Feature flags are table stakes in software
  engineering. In agent systems they're almost nonexistent — prompts, tool
  access, model selection, memory policy, and autonomy level all change globally
  the moment you ship. That's why your "small" prompt tweak just broke 12% of
  your users and you found out from a Discord screenshot. This talk walks the
  feature-flag patterns agents specifically need — beyond the standard
  boolean-toggle. We'll cover segment-targeted prompt variants, per-tool access
  flags, model-routing flags, autonomy-level flags (suggest vs. auto-approve vs.
  auto-execute), memory-policy flags, and the kill switch every agent system
  should have on day one but almost none do. Speakers: - Sachin Gupta: Sachin
  Gupta is a Staff Software Engineer with 15+ years building backend platforms
  at internet scale, currently focused on the runtime trust boundaries that LLM
  coding agents blur and the creator of HeapLens, a Java heap analyzer extension
  used in 50+ countries. LinkedIn: https://www.linkedin.com/in/guptasachin1/
  GitHub: https://github.com/sachinkg12
canonical: 'https://liquidgold.cc/watch/zU4EagB311U'
markdown: 'https://liquidgold.cc/watch/zU4EagB311U.md'
llms: 'https://liquidgold.cc/watch/zU4EagB311U/llms.txt'
parent: 'https://liquidgold.cc/channel/aidotengineer.md'
channel: aidotengineer
updated: '2026-10-24T07:34:50.288Z'
language: null
children: []
videoId: zU4EagB311U
durationSeconds: 1157
publishedAt: '2026-07-18T07:00:06.000Z'
searchChunks:
  - ordinal: 0
    text: null
    startSeconds: 12
    endSeconds: 92
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=12'
    heldReason: rights
  - ordinal: 1
    text: null
    startSeconds: 92
    endSeconds: 103
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=92'
    heldReason: rights
  - ordinal: 2
    text: null
    startSeconds: 103
    endSeconds: 222
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=103'
    heldReason: rights
  - ordinal: 3
    text: null
    startSeconds: 103
    endSeconds: 222
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=103'
    heldReason: rights
  - ordinal: 4
    text: null
    startSeconds: 222
    endSeconds: 233
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=222'
    heldReason: rights
  - ordinal: 5
    text: null
    startSeconds: 233
    endSeconds: 312
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=233'
    heldReason: rights
  - ordinal: 6
    text: null
    startSeconds: 324
    endSeconds: 418
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=324'
    heldReason: rights
  - ordinal: 7
    text: null
    startSeconds: 418
    endSeconds: 428
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=418'
    heldReason: rights
  - ordinal: 8
    text: null
    startSeconds: 428
    endSeconds: 565
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=428'
    heldReason: rights
  - ordinal: 9
    text: null
    startSeconds: 428
    endSeconds: 565
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=428'
    heldReason: rights
  - ordinal: 10
    text: null
    startSeconds: 565
    endSeconds: 575
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=565'
    heldReason: rights
  - ordinal: 11
    text: null
    startSeconds: 575
    endSeconds: 657
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=575'
    heldReason: rights
  - ordinal: 12
    text: null
    startSeconds: 765
    endSeconds: 776
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=765'
    heldReason: rights
  - ordinal: 13
    text: null
    startSeconds: 776
    endSeconds: 910
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=776'
    heldReason: rights
  - ordinal: 14
    text: null
    startSeconds: 776
    endSeconds: 910
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=776'
    heldReason: rights
  - ordinal: 15
    text: null
    startSeconds: 776
    endSeconds: 910
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=776'
    heldReason: rights
  - ordinal: 16
    text: null
    startSeconds: 921
    endSeconds: 1000
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=921'
    heldReason: rights
  - ordinal: 17
    text: null
    startSeconds: 1000
    endSeconds: 1012
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=1000'
    heldReason: rights
  - ordinal: 18
    text: null
    startSeconds: 1012
    endSeconds: 1092
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=1012'
    heldReason: rights
  - ordinal: 19
    text: null
    startSeconds: 1092
    endSeconds: 1104
    watch: 'https://liquidgold.cc/watch/zU4EagB311U?t=1092'
    heldReason: rights
---


# Agents Need Feature Flags - Sachin Gupta

Most AI teams ship behavior changes to 100% of users on every deploy — no canary, no segment, no kill switch. Web teams stopped doing this in 2012. AI teams are about to learn why. Feature flags are table stakes in software engineering. In agent systems they're almost nonexistent — prompts, tool access, model selection, memory policy, and autonomy level all change globally the moment you ship. That's why your "small" prompt tweak just broke 12% of your users and you found out from a Discord screenshot. This talk walks the feature-flag patterns agents specifically need — beyond the standard boolean-toggle. We'll cover segment-targeted prompt variants, per-tool access flags, model-routing flags, autonomy-level flags (suggest vs. auto-approve vs. auto-execute), memory-policy flags, and the kill switch every agent system should have on day one but almost none do. Speakers: - Sachin Gupta: Sachin Gupta is a Staff Software Engineer with 15+ years building backend platforms at internet scale, currently focused on the runtime trust boundaries that LLM coding agents blur and the creator of HeapLens, a Java heap analyzer extension used in 50+ countries. LinkedIn: https://www.linkedin.com/in/guptasachin1/ GitHub: https://github.com/sachinkg12

Canonical: https://liquidgold.cc/watch/zU4EagB311U · 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.
