Talk to an agent from your own application

POST /api/v1/chats/with-message

Creates the conversation and answers the first message in one call. The response is an SSE stream, not JSON — the first event carries the chat_id you need in order to continue.

curl -N -X POST https://chatly.analytos.ai/api/v1/chats/with-message \
  -H "Content-Type: application/json" \
  -d '{"chatbot_id": "YOUR_AGENT_ID", "content": "What are your opening hours?"}'

-N matters: without it curl buffers the response and the stream looks like it arrives all at once at the end.

POST /api/v1/chats/{chat_id}/messages/stream

Sends the next message in an existing conversation. Use the chat_id from the chat_created event. Earlier turns are already in the agent's context — send only the new message.

curl -N -X POST https://chatly.analytos.ai/api/v1/chats/CHAT_ID/messages/stream \
  -H "Content-Type: application/json" \
  -d '{"content": "And on weekends?"}'

Every line is data: followed by one JSON object with a type. Read token and done; the rest are there if you want to show progress, and can be ignored safely. Treat an unknown type as something to skip — more may be added.

data: {"type": "chat_created", "chat_id": "4cff03c1-..."}
data: {"type": "status", "step": "classifying_intent"}
data: {"type": "sources", "sources": [{"title": "handbook.md"}]}
data: {"type": "token", "content": "We are open "}
data: {"type": "token", "content": "9 to 5, Monday to Friday."}
data: {"type": "done", "message_id": "8f1c...", "chat_title": "Opening hours"}

Reading the stream in JavaScript:

const res = await fetch("https://chatly.analytos.ai/api/v1/chats/with-message", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ chatbot_id: AGENT_ID, content: "Hello" }),
})

const reader = res.body.getReader()
const decoder = new TextDecoder()
let buffer = ""
let answer = ""

while (true) {
  const { value, done } = await reader.read()
  if (done) break
  buffer += decoder.decode(value, { stream: true })

  // Events are newline-delimited; the last piece may be incomplete.
  const lines = buffer.split("\n")
  buffer = lines.pop() ?? ""

  for (const line of lines) {
    if (!line.startsWith("data: ")) continue
    const event = JSON.parse(line.slice(6))
    if (event.type === "token") answer += event.content
    if (event.type === "done") return answer
    if (event.type === "error") throw new Error(event.message)
  }
}

GET /api/v1/chatbots/{agent_id}/widget-config

The agent's public presentation — display name, greeting, avatar, colours and which features are switched on. Use it to make your own interface match what you set under Appearance. It returns only what a visitor is meant to see; instructions, knowledge and credentials are never included.

curl https://chatly.analytos.ai/api/v1/chatbots/YOUR_AGENT_ID/widget-config

Failures before the stream starts are ordinary JSON with a detail. Once the stream has started, a failure arrives as an error event instead, and no done follows it.

If you want the ready-made chat bubble rather than your own interface, you do not need this API at all. Paste the snippet from your agent's Publish page:

<script
  src="https://chatly.analytos.ai/chatly-widget.js"
  data-chatbot-id="YOUR_AGENT_ID"
  defer
></script>

The widget handles the streaming, the appearance settings and the conversation history for you.

These two endpoints cover talking to an agent. Creating agents, uploading knowledge, reading conversations and managing connected apps all require a signed-in session and are not part of this public surface — they are done from the dashboard.

An agent only uses connected accounts such as Gmail on behalf of the person who connected them, in the dashboard. A conversation started through this API is an anonymous visitor and never reaches anyone's personal connections.