An AI coworker the whole workspace can use in Slack or Microsoft Teams, with nothing to host.
- No gateway or server to run
- Starter costs US$25 a month per workspace
- Hosted in Australia; 3,000+ tools over MCP
Toby is an OpenClaw alternative for teams that want an AI coworker without setting up and running a gateway themselves. You add it to Slack or Microsoft Teams, and it is ready in about two minutes.
Start with 2,000 free credits, no card needed. US$25 a month after that.
An AI coworker the whole workspace can use in Slack or Microsoft Teams, with nothing to host.
OpenClaw's description: "Your assistant, on your devices, in your chats." [1]
OpenClaw is an open-source AI assistant released under the MIT licence and stewarded by the OpenClaw Foundation, an independent non-profit [1][2]. It runs as a single gateway process on a laptop or a shared team server. People reach it from chat apps such as Slack, Microsoft Teams, WhatsApp, Telegram, Discord, Signal and iMessage [1].
For a team, someone installs that gateway on a host that stays on, such as a small VPS or an office Mac, connects the team chat and keeps it running [3]. The project has no paid tier, hosted service or token of its own [1][2].
Toby joins your Slack or Microsoft Teams workspace as an AI coworker. Someone in a channel asks for a job; Toby does it in the tools you've connected, such as Xero, HubSpot, Gmail or Deputy, then replies in that thread. Nobody on your side looks after a server.
Neither locks you into one model vendor. OpenClaw treats models as plugins you can swap, hosted or local [1]. Toby isn't tied to any single vendor or model either, and it reaches 3,000+ tools. The difference is that Toby comes ready to use.
OpenClaw suits teams that are happy to run their own gateway. Toby differs in three ways.
Toby is ready in about two minutes: install it in Slack or Teams, choose its channels, connect your tools. No machine of yours has to stay switched on. OpenClaw runs as a gateway on a laptop or a server you manage [1][3]. Its installer sets up the Node.js runtime it needs, and there are desktop apps for Windows and macOS [1][4].
Four jobs teams ask Toby for, shown as they run in a channel thread.
A short read at 7:30 covering what is on today and anything that moved overnight.
Toby works out who is late, drafts the follow-ups for you to approve, and tells you which accounts it left alone.
Before a call, Toby writes up the account so you are not reading the CRM in the car park.
It sorts whatever came in overnight and drafts the replies you would have written anyway.
The facts an OpenClaw review would check, next to Toby's.
| What matters | Toby | |
|---|---|---|
| How you get it | A ready-to-use service; nothing to install or host | |
| Where you work with it | Built for Slack and Microsoft Teams | Slack, Microsoft Teams, WhatsApp, Telegram, Discord, Signal, iMessage and 20+ more, plus desktop and mobile apps[1] |
| Model | Not locked to a single vendor or model, and nothing to host | |
| Price to start | First 2,000 credits free (no card), then Starter at US$25 a month per workspace | |
| Setup | About two minutes from install to the first job | |
| Connected tools | 3,000+ tools over MCP; you approve each one by name | |
| Oversight | Shows Approve or Reject before it sends or writes anything | |
| Scheduled work | Repeats jobs on a schedule you choose | Built-in scheduler that can deliver results to a chat channel[8] |
| Hosting | Hosted in Australia |
Your gateway can stay where it is. Give Toby one job the whole team shares.
It takes about two minutes. You pick the channels Toby may read, and private channels and teammates' DMs stay out of reach.
Choose from 3,000+ tools over MCP, such as Xero, Outlook or HubSpot. Toby can't use any of them until you approve it by name.
Mention @Toby with the job and read what comes back in the thread. Once it looks right, tell Toby which day and time to run it again.
Yes, if you would rather not run a gateway yourself. OpenClaw works in Slack and Microsoft Teams through a gateway you set up and keep running [1][3]. For Teams, that includes registering a bot and giving Teams an address it can reach [6]. Toby is a hosted service. Install it in Slack or Teams, pick channels, connect tools, and it is working in about two minutes. Everyone in the workspace shares one plan; Starter is US$25 a month.
Toby can run on frontier open models, which cost a fraction of what closed model labs charge. Toby's first 2,000 credits are free, once. After that, one plan covers everyone: US$25 a month for 20,000 credits on Starter, or US$50 a month for 40,000 on Plus (see pricing). OpenClaw software is free, and the project has no paid tier, hosted service or token [1][2]. You pay for the hardware it runs on and for model access, using a provider API key or an existing subscription login [5].
No. Toby is hosted for you, so your team has no install to maintain and nothing to keep online. OpenClaw's team setup starts with a host for the gateway that stays on, such as a small VPS or an office Mac, with OpenClaw installed and onboarded on it [3]. To watch Toby run a real job first, book a demo.
Work that ends with something done in a business tool. Someone asks in Slack or Teams; Toby plans the steps, completes them in your connected tools and answers in the same thread with what it did. Examples: chasing unpaid Xero invoices, a one-page brief before a client meeting, or Monday's sales numbers. Put a job on a schedule and it runs again without anyone asking.
Over 3,000, through MCP. Common first picks are Xero, HubSpot, Gmail, Outlook, Notion, Google Drive, Salesforce, Jira and Deputy. Toby uses none of them until you approve it by name, and the dashboard can take any of them away later. OpenClaw also connects to MCP servers, alongside its own tools, skills and plugins [1][7].
In Australia. Toby sees only the Slack or Teams channels you add it to. From the dashboard you can turn off one channel, one tool or Toby as a whole whenever you need. With OpenClaw, you choose where it runs: your own laptop, an office machine or a server you manage, using hosted or local models [1][2][3].
Try it in one channel with one tool, and see what comes back.
Start with 2,000 free credits, no card needed. US$25 a month after that.