A remote control for your AI team.
OneTeam gives your AI agents their own private cloud computer, then hands you one console to watch and steer them. Every screen shows the live truth from your own server — never a stale copy. This guide walks you from sign-up to every tab.
New here? Read How OneTeam works and Setup & onboarding first — the rest of the guide is one section per tab, in the same order they appear in your sidebar.
How OneTeam works
When you sign up, we give your AI team a private cloud computer (a VPS — a server that's yours alone). On it we install OpenClaw, the software that actually runs your agents, plus a small companion program of ours called tc-bridge.
The website you log into — this console — is a remote control. It stores almost nothing itself. When you open a screen, the console asks tc-bridge on your computer “what's happening right now?”. tc-bridge reads the real files and running programs on that machine and sends the answer straight back. That's why what you see is always live.
What runs where
- Your VPS — runs the agents, holds all their data, files and memory.
- OpenClaw — the agent engine on your VPS.
- This console — reads from your VPS on demand; steers it when you make a change.
Where the numbers come from
- Billing runs through Stripe.
- AI usage & cost is metered by a gateway called sub2api.
- The cloud computer itself is provisioned on Hetzner.
Your data stays on your server. The console is a window, not a warehouse. Only your account can see your VPS — separation is enforced in the database, not just the app.
Setup & onboarding
OneTeam has two different doors. The first two screens are the same for everyone — after that, Personal and Team split apart. Find your path below and follow only that track.
Which path is yours?
Personal — $30/mo
- Self-serve, card first, fully automatic.
- 1 AI agent · 14-day free trial · $20 AI credit.
- Your computer builds itself while you wait.
Choose Personal if you're one person who wants an AI assistant running for you.
Team — $1,000/mo
- Request first — no card up front.
- 6 AI agents · all channels · multi-agent orchestration.
- We set it up with you after a short chat.
Choose Team if a company or group needs several agents working together.
Shared start — the same for both
Everyone begins here: the front door, then one short form for your email and password.
- Name your workspace & pick Personal. Use your own name — this is your private space, not a company. Then “Continue to payment”.
A personal workspace name (e.g. “Alex Rivera”) with the Personal plan selected. - Add your card. You land on Stripe's secure checkout. The 14-day free trial starts now — you're not charged until it ends, but a card is required to begin.
Stripe checkout: $30/mo, 14-day trial, $20 AI credit. Card details go to Stripe, never to OneTeam. - We set up your AI account. Automatic — we create your usage account and drop in the $20 trial credit. Nothing for you to do.
- Your computer builds itself. We provision your private VPS and install everything. A live checklist ticks off eight steps.
The real setup checklist — steps go green one by one. Safe to leave and come back to. - You're live. When your server reports healthy, you're sent straight to your Dashboard.
- Name your organisation & pick Team. Use your company name. Team's button says “Continue” — no card yet.
A company workspace name (e.g. “Northwind Labs”) with the Team plan selected. - Have the discovery conversation. Instead of paying, you chat with our AI. It asks a handful of questions about your business, your busiest tasks and how you reach clients — one at a time.
The full discovery chat — a natural back-and-forth, no forms. - Get your recommended team. The moment it has enough, it proposes a tailored team — each agent with a role, description and skills, plus an estimated size and cost. Review it, then “Accept & continue”.
Your deployment plan — the agents we recommend for how you work. - We build it — track the status. Your request goes to a OneTeam specialist who provisions everything by hand. This status page shows exactly where you are.
The setup tracker — received → reviewing → building → testing → ready. - Complete payment & launch. When your team is ready you get the go-ahead here (and by email). Hit Complete payment & launch to finish on Stripe's secure checkout — the same card step as Personal.
“Your team is live — complete payment to start.” The pay CTA plus your agent line-up. - You're live. You land on your Dashboard with your full team ready.
Setup usually takes a few minutes. The checklist is safe to leave and return to — every step is resumable, so a refresh or a closed tab never breaks it.
Dashboard
Your home screen — the one-glance answer to “is my team OK today?”
What you see: how active your agents have been today, how many agents you have, today's spend, and your remaining balance — all on one screen.
How it works: the console gathers today's activity totals and your agent roster from OpenClaw on your VPS, pulls spend from the usage gateway, and reads your balance from billing. It's read-only — nothing here changes your server.
Office
A living picture of who's working right now — the fun way to feel your team at a glance.
What you see: a playful 3D “office floor” where each agent is a desk. Busy agents look busy, idle ones look idle, and your roadmap sits on the side.
How it works: every few seconds tc-bridge checks each agent's recent activity and pushes the status live — active (worked in the last 5 minutes), idle, or offline. The roadmap panel reads your workspace's roadmap.md. Read-only.
Activity
The record of what your agents actually did — the “what happened while I was away” feed.
What you see: a feed of events — scheduled runs, model changes — with charts for today's count, success vs. failure, and a heatmap of when things happen. Filter and load more.
How it works: the console merges two sources on your VPS — a live buffer of in-the-moment events, and the full history of scheduled-job runs reconstructed from the run logs on disk.
Agents
Your team roster — and the one place you change how an agent behaves: its model, how much it's trusted, and which chat apps it answers on.
What you see: every AI agent, the model it runs, whether it's online, and how it's wired to messaging. The roster and status are read live from OpenClaw on your VPS.
Inside an agent
Click an agent to open its detail page. It has three tabs down the left — Overview, General, and Channel Config. Every change here writes straight to your live VPS and takes effect on the agent's next run — no restart.
Overview — model & trust
What it does: the agent's identity and its two most important controls.
- Model — the AI brain the agent thinks with. Pick one from the dropdown and hit Save; it applies to new sessions immediately.
- Trust Level — how much the agent can do without asking you first:
- Supervised — approves every file change with you before acting.
- Trusted — auto-approves small changes, asks about big ones.
- Autonomous — full autonomy, no approvals.
General — subagents & fallback models
What it does: settings that apply across every channel.
- Spawnable Subagents — which other agents on your VPS this one may hand tasks to.
- Fallback Models — an ordered backup list; if the primary model is unavailable, OneTeam tries these in priority order so the agent keeps working.
Channel Config — connect messaging
What it does: connect the agent to Telegram, WhatsApp, or WeChat so it can send and receive messages. Each channel is its own tab.
Connect Telegram
- Open Channel Config → Telegram and click Connect.
- Provide your Telegram bot token when prompted.
- The status flips from not paired to paired. Use Disconnect to unlink.
Connect WhatsApp
- Open the WhatsApp tab and click Connect.
- Scan the QR code with WhatsApp on your phone (Linked devices).
- Once paired it shows Paired and starts syncing.
Second Session is an extra WhatsApp connection with no allowlist of its own — it can message any number and feeds inbox monitoring.
Access control. You can limit which Telegram users/groups or phone numbers an agent will respond to — so it only talks to the people you choose.
Sessions
Every conversation and run your agents have had — with full transcripts you can read back.
What you see: a list of sessions. Click one to read the full transcript, see how many tokens it used, and force-stop one that's stuck.
How it works: the list and each transcript are read directly from your agent's session files on the VPS. Killing a stuck session removes it on the server.
Stop a stuck run
- Open Sessions and find the run that isn't finishing.
- Open it to confirm it's the right one, then choose Kill session.
- The run is stopped on your VPS and drops off the list.
Cron Jobs
Your agents' scheduled tasks — “every morning, summarize the inbox” and the like.
What you see: each scheduled job with its schedule, last run, next run and run history. Toggle a job on/off, run it now, or delete it.
How it works: the job list and history are read from your VPS. Run-now and enable/disable go through OpenClaw's real scheduler, so they behave exactly as a scheduled run would.
Run a job now
- Open Cron Jobs and find the job.
- Click Run now. It executes immediately on your VPS through the real scheduler.
- Watch the outcome appear in that job's run history.
Memory
Your agents' long-term notes — the facts they should remember between conversations.
What you see: each agent's memory files — things like “the client prefers X.” Browse and edit them directly.
How it works: the console reads your agent's MEMORY.md and everything in its memory folder. Access is deliberately limited to just those files, so editing memory here can't touch anything else on the server.
Skills
Skills are reusable abilities you can teach your agents — and this is the builder where you create and approve them.
What you see: a three-column Skill Builder — your categorized skills plus a review queue on the left, a chat to author changes in the middle, and a live draft preview on the right. Version history is tracked; publishing is gated behind your approval.
How it works: skills come from three places on your VPS — OpenClaw's built-ins (read-only), your workspace, and managed skills. Every edit is saved to a version history you can revert. When you or one of your agents proposes a change, it lands in Needs your review until you approve it.
Author and publish a skill change
- Open Skills and pick a skill (or start a new one).
- In the chat column, describe the change you want; a draft is written and shown in the preview.
- When it looks right, approve it from the review queue to publish it live to your agents.
Costs
What your AI usage actually costs — the answer to “where is my credit going?”
What you see: your AI usage over time (today, this week, and so on) in both tokens and dollars, with a range selector.
How it works: Costs is the one screen that doesn't read your VPS. The console asks the usage gateway (sub2api) for your account's metered usage. Your keys are never exposed to the browser.
Billing & balance
Your subscription, your remaining credit, and top-ups — all handled through Stripe.
What you see: your current plan and trial status, your usage balance, and buttons to top up, change plan, or manage payment details in Stripe's secure portal.
How it works: subscriptions and payments run through Stripe; your usage credit is a running ledger (trial grant, top-ups, refunds). When a top-up succeeds, the credit is pushed to your AI account automatically.
Low balance? You can set up an automatic top-up so your agents never stop mid-task. Card details live only with Stripe — never in this console.
Files
A full file browser into your agents' data folder — for when you need to look under the hood.
What you see: everything in your agents' data directory — agents, schedules, logs, workspace. Read, edit, upload, make folders, rename, delete.
How it works: the browser reads and writes files directly on your VPS, confined to the OpenClaw data folder so you can't wander into the rest of the machine. It's powerful — changes are immediate and real.
System
Your server's vital signs — and an auto-detected list of problems worth fixing.
What you see: CPU, memory, disk and network usage; which services are running; firewall and network status; your OpenClaw version; plus an Issues panel that flags known problems with a suggested fix.
How it works: tc-bridge reads these straight from the operating system on your VPS. The Issues panel scans your run logs and error logs against a set of known trouble patterns (auth failures, rate limits, disk full…) and attaches a fix suggestion to each.
Terminal
A safe command line into your server, right in the browser — no SSH setup needed.
What you see: a terminal you can type commands into and watch output stream back.
How it works: your command is checked against a safety allowlist before it runs. Read-style commands (ls, cat, grep, ps, systemctl, git, openclaw, tailscale) are allowed; anything destructive (rm, sudo, kill, redirects, curl/wget) is blocked. Safe commands run on your VPS and stream output back live.
Live Logs
Watch your server's logs scroll by in real time — the fastest way to see what's happening right now.
What you see: a live tail of a chosen log — gateway, errors, or commands — updating as new lines are written.
How it works: tc-bridge follows the log file on your VPS (like tail -f) and streams each new line to your browser. Read-only.
Support
Your help desk — open a ticket, watch the thread, and get answers by email.
What you see: your support tickets and their conversation threads. Open a new ticket, reply to an existing one, and track its status.
How it works: tickets live in our system, not on your VPS. Every reply from our team is also delivered to your email. A resolved ticket automatically reopens if you reply again.
Open a support ticket
- Open Support and choose New ticket.
- Pick a category (payment, account, technical, other), write your message, and submit.
- We're notified instantly; replies arrive here and in your email inbox.
Settings
Everything about your account, money and workspace — pinned to the bottom of the dock, split into six panels.
Settings doesn't touch your VPS; it reads your account, subscription and billing from OneTeam and Stripe. It has its own left sub-menu:
Profile
What you see: Your display name (editable, with Save), your email address, and a Change password action.
What it does: Rename yourself, and trigger a password change — we email you a secure reset link rather than taking a new password in the form.
Policy. Your email is your login identity and is shared with your OpenClaw account on the VPS. Password changes always go through an emailed link.
Billing
What you see: Your subscription (Personal $30/mo or Team $1,000/mo plan cards), a trial countdown bar with the exact dates, your AI token credit (balance, usage breakdown, top-up), and Auto Top-Up.
What it does: Subscribe, upgrade or downgrade a plan, cancel or reactivate, update your payment method (opens Stripe), top up token credit, and switch on automatic top-ups so agents never stall on an empty balance.
Policy. The trial runs 14 days and needs a card to start. Cancelling keeps access until the paid period ends. A failed charge moves you to past due; card details live only with Stripe.
Sponsorship Paid plans only
What you see: The Sponsored members panel — pick up the bill for other people's OneTeam subscriptions from your own paid plan.
What it does: Invite members you want to sponsor and cover their subscription. On a trial it's gated behind a “Subscribe to a paid plan to sponsor other members” prompt.
Policy. You must be on a paid plan to sponsor. Each sponsored member keeps their own login, agents and data — only who pays moves to you.
Invoices
What you see: Your billing history — each invoice's number, date, type, status (Paid), and amount, with PDF download and View.
What it does: Download any invoice as a PDF for your records or open it in Stripe.
Policy. Read-only. Invoices are generated and hosted by Stripe; nothing here is editable.
Workspace
What you see: Your organisation name (editable, with Save), your agent limit, and your VPS region.
What it does: Rename your workspace. Agent limit and VPS region are shown for reference.
Policy. The agent limit comes from your plan (Personal = 1, Team = 6). The VPS region is fixed when your server is first provisioned and can't be changed here.
Danger Zone Owners & admins only
What you see: The Delete Workspace action, kept apart from everything else and shown only to owners and admins.
What it does: Start deletion of the whole workspace. It requires an explicit confirm.
Policy. Deletion runs after a 7-day grace period, and billing is paused immediately. Within the grace window it can still be reversed; after it, the workspace and its server are destroyed permanently.
