--- schema: 1 type: video title: 'AI Coding Agents Are Breaking Big Codebases — Dan Adler, Sourcegraph' description: >- Claude Code can make the change. But what if you have 90,000 repositories to make it in? Dan Adler, CEO of Sourcegraph, makes the case for the people who own the huge, decades-old codebases that run banks, cars and airlines. Coding agents are producing a tidal wave of code, and those codebases are starting to decay through duplicated code, drifting standards, brittle dependencies and new vulnerabilities. Agents understand code by searching, but you can't grep what you can't see. Dan argues that large-scale code visibility is infrastructure, and introduces Sourcegraph's Agentic Batch Changes for making changes across thousands of repos from a single prompt. In this talk: • Why most software engineers work in large, long-lived codebases • How AI-generated code accelerates codebase decay • Why context, not model quality, is the bottleneck at enterprise scale • Making and auditing changes across thousands of repos SPEAKER Dan Adler, CEO, Sourcegraph LinkedIn: https://www.linkedin.com/in/danielnealadler/ X: https://x.com/DanielNealAdler LINKS Sourcegraph: https://sourcegraph.com Introducing Agentic Batch Changes: https://sourcegraph.com/blog/introducing-agentic-batch-changes CHAPTERS 0:00 Intro 0:12 Owning a large codebase 0:37 The software that runs the world 2:02 A tidal wave of AI-generated code 2:57 Codebases are starting to decay 3:57 Owning a codebase is harder than ever 4:31 "I don't know what this code does" 5:11 Codebase owners deserve better 6:06 A call to dev tools builders 6:31 "I have 90,000 repositories" 7:21 Agents love to search 7:51 You can't grep what you can't see 8:21 There will be much more code 8:56 The code graph 9:21 Agentic Batch Changes 10:21 Case study: Mercari 11:11 How many repos do you really have? Recorded at the AI Engineer World's Fair 2026 in San Francisco. Subscribe for more talks from the engineers building with AI. AI Engineer: https://ai.engineer YouTube: https://www.youtube.com/@aiDotEngineer X: https://x.com/aiDotEngineer LinkedIn: https://www.linkedin.com/company/aidotengineer/ #AICoding #DevTools #AIEngineer canonical: 'https://liquidgold.cc/watch/Bdrs3uAX0_M' markdown: 'https://liquidgold.cc/watch/Bdrs3uAX0_M.md' llms: 'https://liquidgold.cc/watch/Bdrs3uAX0_M/llms.txt' parent: 'https://liquidgold.cc/channel/aidotengineer.md' channel: aidotengineer updated: '2026-10-23T22:19:59.998Z' language: null children: [] videoId: Bdrs3uAX0_M durationSeconds: 743 publishedAt: '2026-10-04T19:30:20.000Z' searchChunks: - ordinal: 0 text: null startSeconds: 47 endSeconds: 58 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=47' heldReason: rights - ordinal: 1 text: null startSeconds: 58 endSeconds: 125 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=58' heldReason: rights - ordinal: 2 text: null startSeconds: 136 endSeconds: 179 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=136' heldReason: rights - ordinal: 3 text: null startSeconds: 190 endSeconds: 247 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=190' heldReason: rights - ordinal: 4 text: null startSeconds: 190 endSeconds: 247 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=190' heldReason: rights - ordinal: 5 text: null startSeconds: 281 endSeconds: 317 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=281' heldReason: rights - ordinal: 6 text: null startSeconds: 327 endSeconds: 371 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=327' heldReason: rights - ordinal: 7 text: null startSeconds: 407 endSeconds: 443 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=407' heldReason: rights - ordinal: 8 text: null startSeconds: 454 endSeconds: 474 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=454' heldReason: rights - ordinal: 9 text: null startSeconds: 474 endSeconds: 485 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=474' heldReason: rights - ordinal: 10 text: null startSeconds: 520 endSeconds: 542 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=520' heldReason: rights - ordinal: 11 text: null startSeconds: 576 endSeconds: 621 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=576' heldReason: rights - ordinal: 12 text: null startSeconds: 576 endSeconds: 621 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=576' heldReason: rights - ordinal: 13 text: null startSeconds: 631 endSeconds: 674 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=631' heldReason: rights - ordinal: 14 text: null startSeconds: 674 endSeconds: 685 watch: 'https://liquidgold.cc/watch/Bdrs3uAX0_M?t=674' heldReason: rights --- # AI Coding Agents Are Breaking Big Codebases — Dan Adler, Sourcegraph Claude Code can make the change. But what if you have 90,000 repositories to make it in? Dan Adler, CEO of Sourcegraph, makes the case for the people who own the huge, decades-old codebases that run banks, cars and airlines. Coding agents are producing a tidal wave of code, and those codebases are starting to decay through duplicated code, drifting standards, brittle dependencies and new vulnerabilities. Agents understand code by searching, but you can't grep what you can't see. Dan argues that large-scale code visibility is infrastructure, and introduces Sourcegraph's Agentic Batch Changes for making changes across thousands of repos from a single prompt. In this talk: • Why most software engineers work in large, long-lived codebases • How AI-generated code accelerates codebase decay • Why context, not model quality, is the bottleneck at enterprise scale • Making and auditing changes across thousands of repos SPEAKER Dan Adler, CEO, Sourcegraph LinkedIn: https://www.linkedin.com/in/danielnealadler/ X: https://x.com/DanielNealAdler LINKS Sourcegraph: https://sourcegraph.com Introducing Agentic Batch Changes: https://sourcegraph.com/blog/introducing-agentic-batch-changes CHAPTERS 0:00 Intro 0:12 Owning a large codebase 0:37 The software that runs the world 2:02 A tidal wave of AI-generated code 2:57 Codebases are starting to decay 3:57 Owning a codebase is harder than ever 4:31 "I don't know what this code does" 5:11 Codebase owners deserve better 6:06 A call to dev tools builders 6:31 "I have 90,000 repositories" 7:21 Agents love to search 7:51 You can't grep what you can't see 8:21 There will be much more code 8:56 The code graph 9:21 Agentic Batch Changes 10:21 Case study: Mercari 11:11 How many repos do you really have? Recorded at the AI Engineer World's Fair 2026 in San Francisco. Subscribe for more talks from the engineers building with AI. AI Engineer: https://ai.engineer YouTube: https://www.youtube.com/@aiDotEngineer X: https://x.com/aiDotEngineer LinkedIn: https://www.linkedin.com/company/aidotengineer/ #AICoding #DevTools #AIEngineer Canonical: https://liquidgold.cc/watch/Bdrs3uAX0_M · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md) > Search chunk text is held until public-source provenance is verified. ## Document chunks --- #### Chunk 0: Document excerpt (0:12 to 2:05) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 1: Document excerpt (2:05 to 2:59) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 2: Document excerpt (2:59 to 4:31) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 3: Document excerpt (4:31 to 5:17) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 4: Document excerpt (5:17 to 6:36) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 5: Document excerpt (6:36 to 7:54) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 6: Document excerpt (7:54 to 9:02) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 7: Document excerpt (9:02 to 10:21) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 8: Document excerpt (10:21 to 11:14) > Held: transcript rights do not permit this chunk's text. --- #### Chunk 9: Document excerpt (11:14 to 12:23) > Held: transcript rights do not permit this chunk's text.