Building the platform for agent coordination
summary
Linear migrates embeddings from PG vector to Coher and rebuilds search with a hybrid index using Turppuffer, then layers a reranking + query‑rewrite pipeline to ship Product Intelligence (related‑issue maps, suggested labels and assignees), natural‑language filters, issue‑from‑video and customer feedback analysis. An agent platform launched with GraphQL APIs, webhooks, identity, audit trails and an SDK; coding agents produce PRs and root‑cause analyses, and internal estimates suggest agents can clear roughly 50% of a backlog in a week.
AI-written summary · about
chapters (15)
- 0:00Introduction to Linear
- 1:28AI journey at Linear
- 3:56Early AI features
- 5:002024 AI turning point
- 6:24Product intelligence v2
- 7:30Customer feedback analysis
- 8:53Daily workspace pulse
- 9:16Issue from video feature
- 9:58Introducing agent platform
- 10:41Codegen coding agent demo
- 11:44Bucket feature flagging agent
- 12:05Charlie bug analysis agent
- 13:50Future agent interfaces
- 14:53Agent platform architecture
- 16:20Best practices for agents
related talks
34mUnlock the full potential of Generative AI with Agent Platforms
01h 02mPlatform Engineering in the age of Generative AI
38mScaling Specialist Knowledge with AI
43mThe AgentVerse - Build fast, Scale smarter
24mFrom Queries to Agents: The Next Era of Data Retrieval on PostgreSQL
31mBuilding Agentic AI Platforms
01h 00mBuilding a Polyglot Multi-Agent AI Architecture
21mPlatforms for Humans and Machines: Engineering for the Age of Agents