Bring your own AI agent.
Connect Claude, ChatGPT, Codex, Claude Code or pi to one Workspace. It reads and drafts there. You still approve everything.
- Research 120 companies87 / 120
- Transcribe 14 sales calls9 / 14
- Read 60 competitor posts60 / 60
- Draft 40 outreach emails18 / 40
- Scan new tendersMon 06:00
- Keeps going after you close the laptop
- Many jobs at once
- Picks up where it stopped
- A spending cap on every run
One command.
Use claude, codex or pi. You approve with a code in any browser, so it works over SSH too.
Run the command
It shows a sign-in code. For any other MCP client, connect other prints the URL and header.Approve and pick a Workspace
Sign in to Xevol, enter the code and choose where the agent works.Ask your agent
It sees Xevol's tools and works in that Workspace.
npx xevol connect claudeOr add it as a connector.
In Claude, ChatGPT or any app that accepts an MCP server URL, add this address. You sign in to Xevol and pick a Workspace.
https://xevol.com/api/mcpYour agent prepares. You decide.
It acts with your rights, in one Workspace. Paid work it starts uses that Workspace's credits.
Your agent can
- Read the Workspace
- Research and draft
- Create and update records
- Start paid work within your credits
Only a person can
- Approve or reject
- Send, publish or delete
- Change billing, members or apps
What it can work with.
Every tool call is checked against your role and your plan, like any request you make.
You can see and remove every agent.
One Workspace per connection
Fixed when you approve. The agent cannot switch.Connected AI agents
Settings lists each agent, when it was last used and the credits it spent this month.Disconnect any time
Remove an agent in Settings and its token stops working.Standard sign-in
OAuth 2.1 with PKCE and dynamic client registration, over streamable HTTP.
Hand the back office to Xevol.
US$299 a month. You approve everything that goes out.