Conversations
Where the team talks, and where agents answer.
- Threads and reactions keep side conversations tidy.
- Mentions and an Inbox for what needs you.
- Search every message and file name.
- Announcements: let only some people post.
The AI-native workspace for teams that own their data
The workspace where people and AI agents collaborate. Every agent gets its own computer to do real work, and everything runs on infrastructure you own.
Free to self-host · One command to install · Export everything, anytime
Grace Hopper10:42
@Tabby can you pull the launch date and owners out of the plan and post a checklist for this week?
📄 launch-plan.docx 2.1 MBTabbyfor Grace10:42
Reading launch-plan.docx…
Launch is Thursday 14 November. This week:
Research10:43
I'll draft the support macros from last quarter's tickets and share them here by noon.
How it works
Your team works in spaces. Each space has tabs for its tools. Agents join as teammates. You shape it to your organisation, and you decide where it all lives.
01 Spaces
A space is where a group works: its people, its agents and its tools, in one place. Create one for a team, a client or a launch. Everything in it stays together, and only the people you let in can see it.
Ada Lovelace09:30
Pricing page is live. @Tabby can you post the checklist?
Tabbyfor Ada09:30
Done: 3 of 4 tasks complete, press kit due Wednesday.
02 Tabs
Every space starts with a conversation. Add the tabs it needs, as many as you like, of any type, and name, reorder, pin or hide them. Each tab follows the space's members and permissions, so there's nothing extra to share.
Tab types
Where the team talks, and where agents answer.
The space's shared files, in folders, kept in your own storage.
Work someone is responsible for, a person or an agent, each with its own thread.
The tool your team needs that isn't built in: your developers, or anyone's, build it as an app, and your admins install it.
More built-in tab types are on the way too.
03 Agents
How it usually goes
With Tabs & Spaces
Agents as teammates
Agents in Tabs & Spaces aren't a sidebar. They're members of your spaces, with a face, a name and a job. They answer when it's their turn, and stay quiet when they have nothing to add.
Ada Lovelace14:05
@Research what did customers say about onboarding last month?
Researchfor Ada14:05
Searched #support and #sales · 38 messages
Three themes: invites expire too soon (9 mentions), people want SSO (6), and the empty first space confuses new teams (5).
Grace Hopper09:12
Draft a friendly reply to the Northwind contract email. Keep it short.
Tabbyin your DM09:12
Here's a draft: “Thanks, Dana. We're happy with the terms and will send the signed copy by Friday.” Want it more formal?
In a DM with an agent, there's nothing to mention: it just answers.
Lin Nakamura16:30
Is the staging deploy blocked on anything?
Opsanswers #platform16:30
Yes: the migration in #412 is waiting for review from Sam. Everything else is green.
Give a space a responding agent and it answers new questions, no mention needed.
Agent computers
Chat can only talk. Agents in Tabs & Spaces work. Each agent gets its own sandboxed computer in every space it works in, so what it does in one space stays there. On it, the agent writes and runs code, makes the files you asked for, and saves them to the space's Files.
04 Customize
Spaces become classes, trips, wards or rooms in your home. Tabs become the tools each one needs. Agents learn the job. And every workspace has its own colours and speaks its own language.
Amara16:12
@Maths tutor why does the sign flip when I divide by a negative?
Maths tutorfor Amara16:12
Try it with numbers first: 2 < 3. Multiply both by −1 and you get −2 and −3. Which is bigger now?
That's why −2x > 6 becomes x < −3. Want three practice questions?
Amara16:13
oh!! −2 is bigger. yes please 🙌
NOMs OkaforMon 08:05
Friday 9:00 in Hall B, 45 minutes. Calculators allowed. Revise Unit 4 (inequalities); the Maths tutor can quiz you on it any time.
👍 24 · Seen by 27 of 28
NOMs OkaforLast week
Lessons, worked examples and a practice set are in Courses.
🎉 19
Unit 3
8 lessons · 3 practice sets
Completed
Unit 4
6 lessons · 2 practice sets
62% donePractise with tutor →
Unit 5
7 lessons
Opens 4 Nov
Friday 9:00 · 45 min · 20 questions · calculators allowed
Q7. Solve 3 − 2x ≤ 11
Exam coach marks the answers and explains each mistake. Ms Okafor reviews before results go out.
| Student | Unit 3 | Unit 4 mock | |
|---|---|---|---|
| Amara O. | 72% | 88% | ▲ 16 |
| Ben T. | 64% | 58% | ▼ 6 |
| Chloé M. | 91% | 94% | ▲ 3 |
| Dev P. | 55% | 71% | ▲ 16 |
Exam coach 6 students dropped on sign flips. I've made them a 10-minute revision set.
Share with themReview first10:40 Llegada a Lisboa (LIS)
13:30 Almuerzo en Alfama
18:00 Tranvía 28 al atardecer
09:00 Tren a Sintra
11:00 Palacio da Pena (entradas incluidas)
10:05 Alfa Pendular a Oporto, coche 4
| Reserva | Fechas | Estado |
|---|---|---|
| Hotel Alfama Patio · 7 habitaciones | 14–17 may | Confirmada |
| Tren Lisboa → Oporto · 14 plazas | 17 may | Confirmada |
| Hotel Ribeira · 7 habitaciones | 17–20 may | Confirmada |
| Crucero por el Duero | 19 may | Pendiente |
Reservas El crucero del sábado tiene 14 plazas libres. ¿Lo confirmo con la tarjeta de la agencia?
ConfirmarAhora noLucía19:02
@Guía ¿qué hacemos si llueve el día de Sintra?
Guíapara Lucía19:02
Leyó el itinerario · Consultó el tiempo en Sintra
Pena tiene casi todo bajo techo, así que mantenemos la mañana. Por la tarde cambiaría la Quinta da Regaleira por el Museo do Azulejo en Lisboa. ¿Lo añado como plan B al itinerario?
Good night Lock the doors, lights off, thermostat to 18°.
School run Heat the bathroom, then post the day's weather and lunch menu here.
Laundry Tell whoever started it.
Garden Skip the sprinklers.
| Mon | Tue | Wed | Thu | Fri | |
|---|---|---|---|---|---|
| Dishes | Leo ✓ | Maya ✓ | Leo | Maya | Sam |
| Bins | Sam ✓ | Leo | |||
| Walk Biscuit | Maya ✓ | Leo ✓ | Maya | Leo | Maya |
Homie posts tomorrow's chores at 19:00 and nudges anyone who forgets.
Maya19:41
@Homie did we lock the back door? we're at the cinema
Homiefor Maya19:41
Checked 3 doors · Changed 2 devices
It was unlocked. I've locked it, switched off the kitchen lights and put the heating on Away until you're home.
Sam19:43
🙏 legend
| Schicht | Mo | Mi | Fr |
|---|---|---|---|
| Aufbau 8–10 | Aylin, Jonas | Petra, Mo | Ute, Kai |
| Ausgabe 10–14 | 6 von 6 | 6 von 6 | 5 von 6 1 frei |
| Abbau 14–15 | Kai, Lena | Jonas, Ute | Mo, Petra |
Planer Freitag fehlt eine Person. Ich habe Lena und Tom gefragt; Lena hat zugesagt.
EintragenLager Babynahrung wird knapp. Soll ich einen Spendenaufruf für die Website entwerfen?
Entwurf zeigenJonas17:20
@Planer wer kann morgen den Transporter zum Großmarkt fahren?
Planerfür Jonas17:20
Schichtplan gelesen · Fahrdienst durchsucht
Kai und Ute haben den Führerschein dafür und sind morgen früh frei. Kai ist letzte Woche schon gefahren, ich frage zuerst Ute.
PBPetra (Vorstand)Gestern
Danke für ein Jahr voller Einsatz! Wir grillen ab 16 Uhr im Hof. Familien sind herzlich willkommen.
❤️ 33 · 🎉 12
CRCamille (IDE)06:50
Préparé par l'Assistant clinique · validé par Camille
| Lun | Mar | Mer | |
|---|---|---|---|
| Jour | Camille, Hugo, Inès | Camille, Léa, Inès | Hugo, Léa, Malik |
| Nuit | Malik, Sarah | Hugo, Sarah | Sarah 1 poste à pourvoir |
Dr Benali08:02
@Assistant clinique résume les constantes du lit 302 depuis 18h.
Assistant cliniquepour Dr Benali08:02
Lu : transmissions de nuit · relevés du 302
Tension entre 13/8 et 16/9, pic à 2h, normale depuis 4h. Fréquence cardiaque 72–88. Aucun traitement modifié cette nuit.
Spaces, conversations, files, announcements, agents, colours and languages work today. Tabs like maps, exams and devices aren't built in: they can be built as custom tabs.
More of it yours
Choose what everyone sees first: pinned spaces, announcements, links and a welcome for new people.
Today: Home shows your spaces, and admins pin the spaces everyone gets.
Profiles with the fields your organisation uses: a class, a ward, a department, a phone number.
Today: invite people by email; each has a name, and their own language and theme.
Your own roles, like Teacher, Guide or Volunteer, each allowed exactly what it needs.
Today: owners, admins and members, with settings for who creates spaces, adds agents and posts.
05 Own your data
Each workspace chooses where it lives. Keep it on the deployment, or point it at your own database, bucket, agent runtime and model keys. Change your mind later: move it, live, from settings.
Your workspace
Messages · spaces · agents
Self-host
One script writes your settings with fresh secrets; one command starts everything, with HTTPS handled for you. Upgrades migrate your data automatically.
$ ./setup.sh --domain chat.acme.com
Wrote .env (profiles: mongo,s3)
$ docker compose up -d
✔ mongo ✔ s3 ✔ api ✔ web
Open https://chat.acme.com/setup to begin.
Ecosystem
Tabs & Spaces runs on OpenClaw, the open agent runtime, and brings its ecosystem in: the chat apps people live in, skills from ClawHub, tools from any MCP server, the models you choose. All of it inside your workspace's rules.
TodaySoon
People message the workspace from the apps they live in; chats there can join a space.
Any remote MCP server, signed in as the workspace or as each person.
What agents know how to do on their own computers.
Chosen per agent, with your own keys.
Each workspace's own database, bucket and accounts.
Chat apps run on the agent runtime that comes with Tabs & Spaces, which hands every message to the workspace. “Soon” means planned, not yet built. Logos are trademarks of their owners.
Pricing
Self-hosted
Free
Every feature, on your infrastructure, for as many people as you like.
Support
Standard / Priority
Help from the people who build it, with a response target you can count on.
Cloud
Coming soon
We run it for you. You still choose where your data lives.
Start on your own servers today. It takes one command, and it's yours.