Skip to main content
SpeakerWeave runs the program side of a conference: call for papers, structured review, decisions, speaker onboarding, content collection, agenda building, and a public program with embeds and feeds. It is open source (MIT) and built to be hosted as-is or adapted behind clear provider boundaries.

​
What’s in the box

CFP & review

Conditional multi-page forms, anonymized review rounds, weighted scorecards, AI triage with human overrides.

Speaker operations

Rosters, magic-link speaker portals, onboarding tasks, versioned content with approvals.

Agenda & publishing

Drag-and-drop grids, live conflict detection, auto-place, public schedule/speakers pages, embeds, JSON + iCal feeds.

Fully agentic

One org-scoped tool layer behind every surface: in-app agent, MCP server and connectors, Slack, and the sw CLI.

​
Built for the future, and fully agentic

Every surface dispatches through the same organization-scoped tool layer, so the agent can run the program wherever you already work — and anything sensitive stops at an approval gate first.

In-app agent

Streaming threads, @-mention any submission or speaker as context, entity badges that navigate the app, and inline approve/deny gates.

MCP server + connectors

Add /mcp to Claude or ChatGPT as a connector over OAuth. Connectors run inbound too, bringing your own MCP servers into the agent.

Slack

Mention the bot or DM it to collaborate with teammates on live program data, in the channel where the work already happens.

sw CLI

pipx install, authenticate with an API token, then sw ask — the same brain from any shell, script, or coding agent.

​
Try it now

The live demo seeds a full conference workspace — no sign-up: speakerweave.com/demo.

​
Where the code lives

Everything documented here ships in the open-source repository, including the reference deployments for Railway and Cloudflare, the Slack app manifest, the MCP server, and the CLI. If you’re pointing a coding agent at the repo, start it at AGENTS.md.