Lovable launches Chats, detailing the agent architecture behind ~500M daily events
AlexandrePesant · x · 2026-09-24
Lovable launched Chats and published a full engineering breakdown of its multi-agent coordination architecture.
Core design: separating what happened, what an agent needs to know, and when it runs —
- Trajectory: an append-only, forkable history from which each agent's model context is built, rather than treating history as the prompt
- Durable inboxes: agents communicate via persistent mailboxes, handling busy agents and out-of-order instructions
- Activations: scheduling machinery for agents to run and pick up work
Scale: on a typical weekday the Trajectory System appends 500M events across 2.6M user turns. Chats uses this foundation to route confirmed work to the agent building your app and stream progress back into one conversation, with tradeoffs documented publicly.
More from coding & agent
- Treating AI video as a production pipeline: building a reusable cinematic producer agent in CREAO — FellMentKE · 2026-09-24
- Free MCP connector serves UN SDG indicators for all 17 goals by country and year — modelcontextprotocol · 2026-09-24
- Gemini MCP server brings search, deep research and image/video generation into Claude — modelcontextprotocol · 2026-09-24
- Chose rustpython-parser over tree-sitter after measuring both on real LLM output — SprayPuzzleheaded533 · 2026-09-24
- More compute made one agent 13x faster, another just 4%: agents' bottlenecks are task-dependent — alex_verem · 2026-09-24
- At least 4 WebKit WebView 'vibe-coded browsers' appeared in 24 hours — what triggered the wave? — uwukko · 2026-09-24