Modules

One operating brain.
Ten modules.

Each module is useful alone. Together they form a loop: your tools and meetings feed the Brain, the Brain powers the agents, the agents execute with your approval — and everything they learn flows back. Students start with Rooms; Founders Lab adds memory and connectors; companies run the whole loop.

The front door

Super Agent

One chat that knows your whole company.

Ask anything — metrics, decisions, documents, what happened in yesterday’s meeting — and get cited answers drawn from your synced tools and Company Brain. When the answer requires doing something, the agent proposes the action and waits for your approval.

  • Cited, trusted answers over your real data
  • Session history — pick any conversation back up
  • Writes captured back into memory automatically
  • Hands off to a Room when you want to build
Canonical memory

Company Brain

The single source of what’s currently true.

Every meeting, chat, document and connector sync feeds a structured, current-state-first memory: goals, metrics, decisions, people, open loops. It resolves conflicts over time, so agents act on today’s truth — not last month’s.

  • Bi-temporal store: current truth + full history
  • Suggestion queue — you approve what becomes canon
  • Scoped per workspace, never shared across companies
  • Powers every other module’s context
Autonomous outcomes

Missions

Give the agent a goal, not a task list.

“Revive our stale leads” becomes a reviewed plan, an execution loop with checkpoints, and a per-action autonomy matrix — the agent researches and drafts freely, but anything that leaves the building waits for your yes.

  • Plan preview & approval before anything runs
  • Five autonomy levels, set per action type
  • Self-evaluation and reflection after each step
  • Durable — missions survive restarts and resume
Attention layer

Radar

The few things that need a decision from you.

One ranked feed merging approvals, blockers, overdue items, metric anomalies and connector health — refreshed from the Brain. Configurable push over email or WhatsApp so the urgent things reach you without opening the app.

  • Top-10 ranked, not an infinite inbox
  • Anomaly + threshold signals from live data
  • Per-item actions: approve, snooze, open source
  • Push channels with quiet hours
Talk or type → software

Rooms & Builder

Real websites, full-stack apps and native mobile apps.

Describe it and watch it build in a live sandbox: multi-page websites, landing pages, forms, decks, data-backed tools — and now complete applications with real backend APIs, plus native React Native mobile apps previewed on your phone.

  • Auto mode: small edits build, big changes get a plan first
  • Click any element to edit exactly that spot
  • The agent tests its own work in a real browser
  • Every version snapshotted — one-click rollback
Structured truth

Meetings

Meetings that update the company, not just a transcript.

Start a live call in any room. The conversation is captured in-browser, decisions are detected as you speak, and the outcomes flow into trackers, the Brain and the approval queue — so the meeting’s work is done when the meeting ends.

  • No bot joins your call — capture runs in the browser
  • Decisions extracted into the operating tracker
  • Voice or chat: same room, same agent
  • Live artifacts take shape while you talk
Your real stack

Connectors

Gmail, Sheets, Drive, Slack, Notion, Jira, GitHub and more.

Connect the tools your company already runs on. Data syncs on a cadence into your workspace where the Brain, Analytics and agents can use it — governed by permissions you grant per chat, per room or workspace-wide.

  • Off by default — agents ask for access in-chat
  • Synced data powers dashboards and answers
  • Per-agent, per-scope permission grants
  • Sync health visible at a glance
BI over synced data

Analytics

Drag-and-drop exploration, or just ask for the chart.

A full visual explorer over everything your connectors sync — build any chart, filter, or dashboard by dragging fields, or describe the chart you want and the agent builds the spec. Boards save and live inside rooms too.

  • Open-ended: any chart, any field, any filter
  • Agent-generated charts from plain English
  • Server-side compute for big datasets
  • Analytics artifacts embed in rooms
You stay in command

Approvals & Control

Agents execute. Humans authorize.

Every external write — an email, a ticket, a sheet update — is proposed as a card you approve or reject. Artifact locks freeze an app’s structure so agents can only touch data. Workspace rules set what runs free and what always asks.

  • Approval cards for every outside-world action
  • Build ↔ Operate lock on finished apps
  • Configurable autonomy rules per workspace
  • Full audit trail of who approved what
Ship it properly

Publishing & Growth

From sandbox to a real, measurable site.

One click publishes to a stable link with a QR code. Bring your own domain with a DNS record. See visitors and referrers, run a security scan before you share, publish any app as a remixable template — and your source is always yours to download.

  • Stable share links + custom domains (beta)
  • Visitor analytics on every published site
  • Security scan + template gallery with remix
  • Source export & GitHub sync — no lock-in

Want the flat list instead? See every feature

Walk in with an idea. Leave with a prototype.

Invite-only beta · First room free