The personal agent, for work
Give every teammate an agent that lives in Slack. Enterprise-grade permissions, out of the box. Script it all with API or MCP.
Use any model. Switch when a better or cheaper one comes out. Your agents keep their memory and instructions.
Don’t make your coworkers sign up for another agent.
Everything happens in Slack, including approvals.
Own your data.
We don’t hold your data hostage. Access full agent traces, memories, and context via API. Understand the full set of system prompts and tool usage.
Permissions you can reason about.
A shared company bot needs access to everything, which makes it hard to say what it should be allowed to touch. In Matrices, each agent acts on behalf of one person and can only see what that person can. That also limits what a prompt injection can reach. You also decide what needs a person’s sign-off, like sending email, and cap what each agent can spend.
- Dana
- Gmail
- Calendar
- Salesforce
- HubSpot
- Slack
- Leo
- GitHub
- Linear
- Jira
- Notion
- Slack
- Drive
- Gmail
- Priya
- Brex
- QuickBooks
- Sheets
Up and running in five minutes.
- 1
Install the Slack app
One admin signs up, names the app, picks its icon, and installs it. It takes about five minutes.
- 2
Each person messages it
The first time someone messages it, their agent introduces itself, helps them connect their accounts, and explains what it’s allowed to do.
curl -X POST api.matrices.com/v1/agents \ -H "Authorization: Bearer $MATRICES_KEY" \ -d '{ "identity": "dana@yourco.com", "profile": { "models": ["anthropic/claude-sonnet*"], "roleIds": ["role_staff"] }, "budget": { "limit": 25, "window": "7d" } }'