Embeddable conferencing · agent-native

Own the room.
Let your AI run it.

confcall is the embeddable video & phone room with an AI agent built in — not a bot that joins someone else's Zoom. Drop in one component, call one API, and the transcript, summary, and action items land in your database.

▹ Web · phone · chat ▹ REST + MCP control plane ▹ Built on LiveKit
AppointmentScreen.tsx embed
// drop the room into your own app
<conference-room
  room="visit-8420"
  agent="notetaker"
  on-artifact={saveToChart} />

// …the agent joins, runs the call,
// and posts structured output back:
await confcall.rooms.artifacts(id)
// → { transcript, summary, actions[] }
live · visit-8420
host
guest
◆ AI agent
+ dial-in

The difference

A bot can join a Zoom call. It's still a guest.

Meeting bots are genuinely capable now — they transcribe, look things up, even write to your CRM. The difference isn't what the agent can do; it's where it lives. A bot works inside a platform you don't control, on a call your users had to leave your product to attend. confcall makes the call a surface of your app, on infra you own.

A bot in someone else's meeting

Capable — but on rented land

It can act, but it's on rented land — their API limits, their bot policies, their pricing. Build on it and you're exposed to rule changes, and to the platform shipping its own agent to compete with you.

  • — Users still leave your product to meet in Zoom
  • — Joins as a visible third-party bot — consent friction, can be blocked or banned
  • — Gets a media feed, not your app's identity, the on-screen record, or room control
  • — Web-meeting-only, plus a bot vendor in your data path
confcall

The room is part of your product

The call renders inside your app and runs on infra you control, so the agent is native to the room — on your platform and your roadmap, not a visitor negotiating for access.

  • + The meeting is a surface of your app — no context switch, no second tool
  • + Agent acts with your identity & context, controls the room, and can drive embedded UI (cobrowse, forms)
  • + Your consent flow, your data path — cleaner for compliance, no third party in the middle
  • + One agent across web, phone, and chat, from a single control plane

See it work

A room with an agent in it — and the notes already done.

Host and guests on video, your AI agent listening as a participant, and the transcript turning into action items that sync to your CRM before anyone hangs up.

UI preview — drop in a recorded walkthrough here for launch.

One control plane

Web, phone, and chat — from one API.

Competitors make you stitch a video SDK to a phone-AI vendor to a chat widget. confcall spans all three channels behind a single control plane for rooms, agents, and artifacts.

01 / web

Browser video rooms

A drop-in <conference-room> web component or hosted embed page. No download, no WebRTC plumbing to maintain.

WebRTC · screen share · recording
02 / phone

PSTN & SIP agents

Bring a number or your Twilio/SIP trunk. The same agent answers the phone, runs an IVR flow, and escalates to video.

Twilio · SIP · IVR builder
03 / chat

Embeddable text agents

The same brain as a chat widget for your site or app — one conversation model across every surface a customer reaches you on.

Widget · web component · API

The retention loop

The call is disposable. The artifact isn't.

Nobody rewatches a recording. The value is the structured result — a summary, the action items, the fields — landing in the tool your team already lives in, the moment the call ends.

01

Embed the room

One component or one POST /rooms. Mint a scoped join token for your user.

02

The agent runs the call

Joins as a participant — transcribes, answers, verifies, keeps the thread.

03

Artifacts land in your system

room.ended → a signed webhook with transcript, summary, and typed action items.

◆

You never touched a media server

Real-time infra, agent pipeline, and storage are ours to run. You get the result.

Built to embed

Primitives, not a walled app.

Everything is a building block you compose into your product — reachable from your frontend, your backend, or an AI agent operating it for you.

<conference-room>

A framework-agnostic web component and hosted embed pages. Props in, lifecycle events out. Ships as a tiny script.

REST control plane

Create rooms, invite agents, mint join tokens, pull artifacts. Clean, versioned, key-scoped per tenant.

MCP server

Expose rooms, agents, and artifacts as tools so an agentic host — a CRM copilot, your own orchestrator — can operate confcall directly.

Signed webhooks

Subscribe to room and agent events. HMAC-signed, retried with backoff, delivered to the endpoint you register.

Pricing

Start free. Then pay for what you use.

Usage-based on participant-minutes — no per-seat tax for embedding, and the rate you see is the rate you pay. Bring your own SFU or telephony if you'd rather.

Developer
  • ✓ A real free tier — build and test before a card
  • ✓ Per participant-minute, billed by the second
  • ✓ Transcripts, summaries & artifacts metered transparently
  • ✓ No seat lock-in; scoped API keys per tenant
  • ✓ Self-host the media plane (LiveKit) when you scale
Get started
from
$0.004
/ participant-minute · example rate

Figures shown are placeholders for the launch page — swap in confcall's published rate and free-tier limit before go-live.

Who builds on confcall

For the people shipping the product around the call.

Small SaaS teams, indie builders, and agencies who need a call — with an agent in it — inside their own app, not another tab their users have to leave.

telehealth in the EHR advisory in the CRM coaching platforms marketplaces field service support escalation AI phone lines

Put an agent in your room.

Spin up an embeddable room with an AI participant and watch the first artifact land — in about an afternoon.