The CRM your
AI voice agent actually runs.
Contacts, pipeline, tickets, inbox — wired into an MCP server so your agent qualifies leads on outbound, resolves support on inbound, and logs every touch alongside your reps.
Inbound + outbound calls with live transcripts and CRM writes.
Drag-and-drop deals, per-stage probability, won / lost tracking.
B2B and B2C, opt-ins, custom fields, unified timeline.
SMS, WhatsApp, chat — one thread per conversation.
Your AI answers, calls back, and writes to the CRM.
Every call the agent takes opens an AgentSession, resolves the caller by phone, and logs a transcript with a summary and outcome. Every action the agent takes goes through the same service layer your reps use — auditable, attributed with a violet AI chip, and idempotent.
- Plivo inbound/outbound with signature verification
- Session tool-log — see every action the agent took
- OpenAI Realtime or Deepgram Voice Agent bridge
One tool registry. Three ways to call it.
Your voice agent, your MCP-connected chat agent, and your external integrations all speak to the same 21 tools — reusing the exact code path your web UI uses. Add a service method once, agents can call it everywhere.
- MCP over stdio (Claude Desktop)
- MCP over HTTP JSON-RPC
- REST /api/v1/* with Bearer keys
Deals move themselves when the agent qualifies them.
Your agent can create deals from calls, move stages when it hears buying signals, and set follow-up tasks — every change lands in the unified timeline with an AI attribution.
- Drag-drop kanban with stage rules
- Won/lost automation on outcome
- Full activity log per deal
Ship your voice agent this week.
Spin up a workspace, generate a Bearer key, wire it to your agent. Free while you build.