Released 2026-06-23. Put AI in your realtime streams — your team's own AI model can answer players in a room, with nothing to host.
Added
- Call AI from a channel. Bind a stream channel to Call AI: write a system prompt and every player message runs your team's own AI model, with the reply published back into the room. No agent, workflow, or code. Set a max-reply-tokens cap and a per-room cooldown to keep credit spend in check, and JSON the model returns arrives as structured data.
ctx.ai.call()in stream code. A stream Code function can now call AI inline —const { text } = await ctx.ai.call(message, { system, max_tokens })— on the same team model, for full control over the prompt and what you publish back.
Notes
- Both run on your team's configured AI and are billed to your team's AI credit pool, with the same balance and monthly limits as every other AI feature.