Added
- Coworkers — create focused chat personas in the sidebar, each with its own memory, colour, and standing instructions; keep them private, share with specific teammates, or open to your whole org
- Developer API — a global Apex API (
lai.LogicAI) to call Logic AI from your own Apex, Flows, and triggers - Usage limits — set monthly USD spend limits per source and per user, with a new per-source and per-user spend breakdown in the admin console
- Metadata tools (admins only) — Logic AI can read how your org is built (Apex, Flows, validation/workflow rules, permission sets, LWC/Aura) to explain your automation
- Smarter memory — memory is now captured automatically and recalled invisibly, with a health score and 👍/👎 feedback that auto-tunes how much Logic AI remembers
- Chats sidebar — one-line chat rows, delete chats, a resizable divider, and smoother loading
Fixed
- Fixed a freeze that could occur when giving a message a 👎
- Removed occasional blank assistant messages from the chat
Improved
- Newer AI models — Logic AI now runs on the latest Claude models: Sonnet 5 (up from Sonnet 4.6) and Opus 4.8 (up from Opus 4.7). Your existing model selections pick up the upgrade automatically — nothing to reconfigure or reinstall
- Longer, more reliable timeouts on long-running requests
- Memory can be rebuilt from a full history scan, migrating older memory into the new system