Managed Agents on SLNG

Managed Agents pair an LLM with SLNG’s STT and TTS models. Create an agent once, then dispatch it to phone calls or web sessions, with shared tools, Custom Code, MCP servers, and template variables.

Voice Agents are server-managed configurations that pair an LLM with SLNG’s STT and TTS models. You create an agent once, then dispatch it to phone calls or web sessions. Agents can use published organisation tools, sandboxed Custom Code, MCP servers, built-in capabilities, and dynamic prompts.

The Voice Agents API is served from https://api.agents.slng.ai under /v1/agents.

Where to go next

This page is the orientation. The detail lives in four places:

Build your first agent

Create, update, duplicate, and delete agents in cURL, JavaScript, and Python.

Configuration & Tools

System prompts, models, shared tool attachments, and template variables.

Calls, telephony & web sessions

Dispatch outbound calls, run batch campaigns, and test with web sessions.

Managed Agents

Build and test agents with the visual Dashboard UI.

Dashboard or API

You can build agents either way, and the two stay in sync. Start in Managed Agents for a UI (and to set up Telephony), or manage everything through the API. Everything in the Dashboard maps to public API endpoints.

If you already have an agent configured the way you want, you can duplicate it from the API or from Agent Infra. Duplication preserves the stored configuration, but it does not copy call history or the inbound connection. After duplicating, reconnect inbound routing if needed, verify the outbound trunk, and test the copy with a web session or test call.

Versioning, portability, and secrets

• Version history and restore — every save creates an immutable version; preview and restore any of them via the versions endpoints.
• Config export and import — download an agent’s canonical config document and recreate it as a new agent.
• Shared tool versions — publish a tool once, then attach an immutable version to any number of agents. Publishing a new version does not change existing agents.
• Vault variables and secrets — keep shared values and credentials in Vault and reference them from agent configuration instead of embedding them.

Agent regions

Voice Agents use logical region values that correspond to the Agent Settings region menu. The selected region controls where the agent session runs and how the agent routes its model requests. For model requests, routing is constrained to the selected compliance area, and the exact deployment can vary based on model availability inside that area.

Agent Settings label	API region value	LiveKit runtime location
Europe	eu-central	Frankfurt, Germany
United States	us-east	Ashburn, Virginia, USA
Asia	ap-south	Mumbai, India

Example agent patterns

Common production-style tool combinations (configured in Configuration & Tools):

• Scheduling assistant: Current Date and Time + End Call + idle nudges. The agent gets accurate local time before it offers appointment slots.
• Support router: Custom Code or API Request + Transfer Call. It resolves routine cases and escalates exceptions.
• Collections follow-up: Voicemail Detection + an automatic API Request on call_end. Every completed call posts a summary to your CRM.
• Restaurant host: Current Date and Time + an availability API Request. It can check live reservation inventory.
• After-hours clinic router: Current Date and Time on call_start + Transfer Call + End Call. Business-hour context reaches the model before the first turn.

Calls and sessions

Endpoint	Purpose
POST /v1/agents/{agent_id}/calls	Dispatch an outbound call (requires telephony)
POST /v1/agents/{agent_id}/web-sessions	Create a web (non-telephony) session
POST /v1/agents/{agent_id}/duplicate	Server-side copy (inbound connection and call history are not copied)
GET /v1/agents/{agent_id}/calls	List calls (paginated)
GET /v1/agents/{agent_id}/calls/{call_id}	Fetch a call record
GET /v1/agents/{agent_id}/calls/export	Export the call list as CSV (same filters as the list endpoint)

For code, see dispatching calls and the agent lifecycle examples. For a friendly transcript view, use the Dashboard calls page.

API reference

See the Voice Agents API for agent lifecycle operations. The Agent resources group in the API Reference covers shared tools, MCP servers, Vault entries, and client models.