Questions

Questions about AI bots and agents in Europe.

The homepage answers the first eight. Here's the rest: agents at once, accounts, data, models, teams, and security.

Free: 30 answers a day, plus 300 trial credits for every other model. No card.

FAQ

Everything else.

Can several chats work at once?

Yes. Each chat keeps working on its own, so start as many as you need while your bots run on their own triggers. Close your laptop and they carry on; you hear when one is done or needs you.

Whose account does a bot use?

Each integration acts as the bot itself, like the GitHub App commenting as chat-with-work[bot], or as the person who lent their account. Only you can lend yours, and its changes ask you first unless you let them run.

More in the help center: Bots

Which AI agents aren't available in Europe?

As of October 25, 2026, Claude Duet is a research preview for the US only, and ChatGPT's dots leave out Pro users in the EEA, Switzerland, and the UK. Chat with Work runs in the EU on every plan, Free included. The alternatives page says where each product runs and on which plans, with sources.

Does Chat with Work copy or index my company data?

No index. When you ask, it searches your tools and reads only what the answer needs. What it reads is cached, encrypted, and purged after 30 days unused. Chats stay until you delete them.

More in the help center: Privacy and EU hosting

Which tools can it connect to?

1,892 tools, from files, mail and calendar, and chat to CRM, finance, legal, design and media, and travel and property. Add any remote MCP server, or your own computers through the open-source Chat with Work Desktop app, for the rest. The full list is on the integrations page.

More in the help center: Connecting services

Can it write code and make files?

Yes. It runs code in a sandbox on our servers, with no network and only the chat's files, to calculate and to build decks, documents, and spreadsheets on your company's templates. On your own computer, it can run commands like your tests in folders you allow, each one approved.

Does it remember things?

It reads instructions you write for yourself, your projects, and your organization, and saves short facts about you that you can see and forget in Settings.

Is Chat with Work an OpenClaw alternative?

Yes, for teams that want bots at work without installing, hosting, and securing an agent themselves. They run on our servers in the EU, each person's accounts stay their own, and changes ask first unless you let them run. OpenClaw fits people who want full control of an open-source agent on their own machine.

Can I connect more than one account of the same tool?

Yes, on most integrations. Connect a workspace per client, or a work and a personal account, and name each one: the assistant knows which is which and searches each one separately. Each account keeps its own access level, so one can stay at Read while another can make changes, each with your approval.

More in the help center: Connecting services

Can my team use it together?

Yes. Put a chat in a project and your teammates on it read the answers live, as they're written, and can ask the next question with their own tools. Guests read only the chats you show them. Each person connects their own accounts, so what your tools return stays with you: teammates see the answer, never your access. Share a bot in a project too: everyone on it reads its conversations, and the people who look after it approve what it waits on.

More in the help center: Projects, roles, and sharing

Can I add my own MCP servers?

Yes. Add any remote HTTPS server in Settings. It starts at Read, offering only its read-only tools. What the server does with your requests is up to whoever runs it.

Can Chat with Work change things in my tools?

Only if you let it. Every tool starts at Read. Switch one to Read and change, and each change asks first, shown exactly as it will run, unless you let a chat's edits or a bot's changes run. Deleting asks unless you allow that one tool, or admins allow "Allow everything" and you turn it on in a chat. Admins can turn changes off for everyone.

More in the help center: Approving changes

Can I choose the AI model?

Yes, per chat: Gemini, GPT, Claude, or the free Qwen model, all in the EU. Self-hosted installs can use any provider, or local models through Ollama or GPUStack.

Can it search the web?

Yes, if you want it to. Web search is off by default: an admin allows it for the organization, then you turn it on in a chat. The searches go to Google for Gemini or Bing for GPT, outside the EU, and each costs 1.4 credits.

Do you train AI models on our data?

No. We don't train models on your data, and the hosted app uses model providers whose business terms say customer data is not used to train their models.

What is the security and compliance status?

The CASA Tier 2 assessment is complete, Google OAuth App Verification is approved for the read-only Drive, Docs, Sheets, and Slides scopes, and the SOC 2 Type I audit process is underway. The security page explains the main controls in plain language.

Back to the first eight

Still wondering? Try it.

Start free, pick a bot, and connect one tool. Changes ask first, unless you let them run.

Free: 30 answers a day, plus 300 trial credits for every other model. No card.