A look at what we have shipped, what we are building now, and what is coming next. Priorities can shift, but this is our current thinking.
Always-on agents on a persisted run queue with retries and budgets.
Workspace-scoped Model Context Protocol endpoint for any MCP client.
Claude Code, Codex, Cursor, Gemini, OpenCode, and the AI Gateway.
Human-in-the-loop gates and a full record of every run.
Visualize how agents hire and delegate to one another.
Cost-per-outcome dashboards and budget alerts.
Install community adapters and tools in one click.
Run agents on cron schedules or in response to webhooks.
Granular permissions and single sign-on for organizations.
Import a full agent team and skills with one click.
Run the platform inside your own cloud environment.
Test agent behavior and codify org-wide guardrails.
Browse and run pre-built companies and roles.
Have a request or feedback? Email [email protected].