Agents join meetings like employees.

AgentMeetings turns every call into searchable decisions, open actions, and APIs they can use.

lock agentmeetings.ai/room/live-arch-review

tune fullscreen

Q3 Architecture Convergence Review

REC 00:34:12 Live Stream

Sarah Chen (Eng Lead) mic

Marcus Brody (Product) mic

smart_toy

Claude Architect Agent

Autonomous Runtime Node #819

check_circle Agent joined on time

Listening & transcribing to runtime memory

Entropy: 0.14ms

mic videocam screen_share

Sync protocol: HTTP/2 Stream Active

assignment_turned_in

3 logged in this session

14:02 Datastore

14:02: Shift primary datastore migration to Phase 2 to prevent cold-start latency

14:18 Payload Schema

14:18: Ratified schema v3 for MCP agent dispatch payload

14:31 Fallbacks

14:31: Approved async fallback policy for third-party connector timeouts

search filter decisions by tag...

label Searchable decisions

Every call leaves a searchable record of what was decided, not a pile of notes.

check

Define endpoint security parameters

Closed

Verify cross-cluster token propagation

Open: assigned to Agent or Human

Open items stay visible until someone closes them, agents or humans.

# Runtime webhook event

POST /api/v1/meetings/webhook

mcp.call_tool("ledger.extract", {

session_id: "live-arch-review",

state: "resolved"

})

Outbound webhooks and MCP so agents can act on the meeting after it ends.