Chat agents are the same idea as voice agents — a prompt, a model, knowledge, tools — running over text instead of audio. Use them for in-product support, lead capture on your website, or WhatsApp/SMS-style conversations through your own gateway.
Creating a chat agent
Create one from Agents → Create Agent by choosing Chat Agent as the mode. Configuration mirrors voice agents — system prompt (single or multi prompt), LLM provider and model, knowledge base attachments, and memory — minus everything audio-specific. The prompt writing guide covers how chat prompts should differ from voice prompts.
Sessions
A conversation with a chat agent is a session: it belongs to one customer (identified by a customer_id you choose), keeps the full message history, and carries metadata you pass in — form fields, account context, prior interaction summaries.
Sessions have a status — active, dormant (quiet for a while), or closed — and a mode:
ai — the agent responds autonomously.support — a human operator is involved.
You drive sessions programmatically with the Chat v1 API: create a session, send customer messages and read the agent’s replies, and inject context mid-conversation (for example, a summary of a phone call that just happened, so the chat picks up where the call left off).
Human-in-the-loop support
Chat agents are built for supervised operation. When a customer asks for a human (or the agent decides it’s out of its depth), the session is flagged as help requested, and operators can step in through support commands:
| Command | What it does |
|---|
| Whisper | Coach the AI privately — the agent incorporates your guidance, the customer never sees it. |
| Tell | Send a message directly to the customer, alongside the AI. |
| Takeover | Switch the session to support mode — you respond, the AI stands down. |
Escalating a chat to a phone call
A chat agent can offer to call the customer: when the customer agrees, OSVI places an outbound call from a voice agent, passing along a summary of the chat so the caller doesn’t repeat themselves. After the call, its summary can be injected back into the chat session.
Reviewing and supervising sessions
Conversations → Chat is the session log, with stat cards (total, active, closed, dormant) and filters for agent, status, mode, customer ID, and help requested — that last one is effectively your support inbox.
Opening a session shows the full message thread with timestamps, the session’s metadata, and its current mode, so an operator has full context before stepping in.
Watch the help requested filter during your first weeks live. The sessions where customers asked for a human are your best source of prompt improvements — each one is a conversation the agent should eventually handle alone.
