Create Chat Session

curl --request POST \
  --url https://api.talkifai.dev/v1/chat/sessions \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api-key>' \
  --data '
{
  "agent_id": "5b710eca-ee67-4c3a-aeb6-8b541f451b40"
}
'

{
  "conversation_id": "chat_5b710eca_user123_1705312800000",
  "session_token": "<string>",
  "agent_name": "Support Agent",
  "greeting": "Hello! How can I help you today?",
  "agent_model": "gpt-4o-mini"
}

POST

chat

sessions

Create Chat Session

curl --request POST \
  --url https://api.talkifai.dev/v1/chat/sessions \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api-key>' \
  --data '
{
  "agent_id": "5b710eca-ee67-4c3a-aeb6-8b541f451b40"
}
'

{
  "conversation_id": "chat_5b710eca_user123_1705312800000",
  "session_token": "<string>",
  "agent_name": "Support Agent",
  "greeting": "Hello! How can I help you today?",
  "agent_model": "gpt-4o-mini"
}

Overview

Creates a new chat session for a text-based AI agent. This endpoint initializes the agent, loads memory context (if enabled), starts billing, and returns a JWT session token for subsequent requests. Authentication: X-API-Key or X-Studio-Token header required Time: ~300-500ms (agent initialization + memory loading)

Request

Headers:

Content-Type: application/json
X-API-Key: tk_live_your_api_key_here
# OR
X-Studio-Token: better_auth_session_token
Origin: https://your-domain.com (optional, for CORS)

Body:

{
  "agent_id": "5b710eca-ee67-4c3a-aeb6-8b541f451b40",
  "metadata": {
    "source": "website_chat",
    "page_url": "https://example.com/support"
  },
  "user_identifier": "customer@example.com"
}

Request Fields

Field	Type	Required	Description
`agent_id`	string	✅	Agent ID (UUID format)
`metadata`	object	❌	Optional metadata for tracking
`user_identifier`	string	❌	End-user email for memory linking

User Identifier (Important)

The user_identifier should be the end-user’s email address. This enables:

Cross-channel memory: Links web chat, voice calls, and telephony conversations
Returning user context: Memory persists across sessions
Personalized responses: Agent remembers past interactions

If not provided, a throwaway UUID is generated and memory cannot be retrieved later.

Response

Status: 200 OK

{
  "conversation_id": "chat_5b710eca_user123_1705312800000",
  "session_token": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...",
  "agent_name": "Support Agent",
  "greeting": "Hello! How can I help you today?",
  "agent_model": "gpt-4o-mini"
}

Response Fields

Field	Type	Description
`conversation_id`	string	Unique session ID (use in all subsequent requests)
`session_token`	string	JWT token for authentication (24h expiry)
`agent_name`	string	Display name of the agent
`greeting`	string	Agent’s opening message (null if user speaks first)
`agent_model`	string	LLM model used (e.g., `gpt-4o-mini`)

Authentication Modes

Mode 1: API Key (External Embeds)

Use Case: Customer-facing chat widgets on external websites Flow:

Your backend calls /v1/chat/sessions with your API key
Your backend receives session_token
Your backend sends session_token to frontend
Frontend uses session_token for all chat requests

Security: API key never exposed to browser

// Your backend server (Node.js/Express)
app.post('/api/chat/start', async (req, res) => {
  const response = await fetch('https://api.talkifai.dev/v1/chat/sessions', {
    method: 'POST',
    headers: {
      'X-API-Key': process.env.TALKIFAI_API_KEY,
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      agent_id: process.env.TALKIFAI_AGENT_ID,
      user_identifier: req.body.userEmail // Optional
    })
  });

  const data = await response.json();
  res.json(data); // session_token is safe to send to frontend
});

Mode 2: Studio Token (Internal Preview)

Use Case: TalkifAI Studio agent testing and preview Flow:

User logged into TalkifAI Studio
Frontend has Better Auth session token
Frontend sends X-Studio-Token header
Session created with user’s active organization

// TalkifAI Studio (Next.js)
const response = await fetch('/v1/chat/sessions', {
  method: 'POST',
  headers: {
    'X-Studio-Token': session.token,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    agent_id: agentId,
    user_identifier: user.email
  })
});

What Happens on Session Creation

Authentication Verified
- API key validated against database
- OR Studio token verified against session table
Agent Loaded
- Agent configuration fetched
- Custom functions loaded
- Subagents loaded (if configured)
- Knowledge bases linked
Memory Context Resolved
- If user_identifier provided: resolve Graphiti memory_id
- Load historical context from knowledge graph
- Inject into system prompt
Credit Check
- Organization credits verified
- Fail-closed: insufficient credits = 402 error
Billing Session Started
- Billing session created in database
- Session type: chat
- Ready for message-based billing
Session Stored
- Session stored in memory (SessionStore)
- Conversation input initialized
- Activity tracking started
Greeting Generated
- Based on agent’s greetingType
- Saved to ChatMessages table
- Returned in response
JWT Token Signed
- Contains: conversation_id, agent_id, organization_id, user_id
- Valid for 24 hours
- Returned as session_token

Error Responses

400 Bad Request

{
  "detail": "Agent ID is required"
}

{
  "detail": "Agent not found or does not belong to your organization"
}

401 Unauthorized

{
  "detail": "Authentication required: provide X-API-Key or X-Studio-Token"
}

{
  "detail": "Invalid or expired studio session"
}

402 Payment Required

{
  "detail": "Insufficient credits. Please add credits to continue."
}

404 Not Found

{
  "detail": "Agent {agent_id} not found or does not belong to your organization"
}

503 Service Unavailable

{
  "detail": "Billing service unavailable: connection timeout"
}

Examples

cURL (API Key)

curl -X POST "https://api.talkifai.dev/v1/chat/sessions" \
  -H "X-API-Key: tk_live_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "5b710eca-ee67-4c3a-aeb6-8b541f451b40",
    "user_identifier": "customer@example.com"
  }'

JavaScript (Backend Proxy)

// server.js (Node.js/Express)
app.post('/api/chat/start', async (req, res) => {
  try {
    const response = await fetch('https://api.talkifai.dev/v1/chat/sessions', {
      method: 'POST',
      headers: {
        'X-API-Key': process.env.TALKIFAI_API_KEY,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({
        agent_id: process.env.TALKIFAI_AGENT_ID,
        user_identifier: req.body.email,
        metadata: {
          source: 'website_chat',
          page: '/support'
        }
      })
    });

    if (!response.ok) {
      const error = await response.json();
      return res.status(response.status).json(error);
    }

    const data = await response.json();
    res.json(data);
  } catch (error) {
    res.status(500).json({ error: 'Failed to start chat' });
  }
});

Next.js API Route

// app/api/chat/start/route.ts
import { NextResponse } from 'next/server';

export async function POST(request: Request) {
  try {
    const body = await request.json();
    
    const response = await fetch(
      `${process.env.TALKIFAI_API_URL}/v1/chat/sessions`,
      {
        method: 'POST',
        headers: {
          'X-API-Key': process.env.TALKIFAI_API_KEY!,
          'Content-Type': 'application/json'
        },
        body: JSON.stringify({
          agent_id: process.env.TALKIFAI_AGENT_ID,
          user_identifier: body.email,
          metadata: body.metadata
        })
      }
    );

    const data = await response.json();
    
    if (!response.ok) {
      return NextResponse.json(data, { status: response.status });
    }

    return NextResponse.json(data);
  } catch (error) {
    return NextResponse.json(
      { error: 'Failed to start chat' },
      { status: 500 }
    );
  }
}

Python (FastAPI)

# main.py
from fastapi import FastAPI, HTTPException
import httpx
import os

app = FastAPI()

@app.post("/api/chat/start")
async def start_chat(email: str, metadata: dict = None):
    async with httpx.AsyncClient() as client:
        response = await client.post(
            "https://api.talkifai.dev/v1/chat/sessions",
            headers={
                "X-API-Key": os.environ["TALKIFAI_API_KEY"],
                "Content-Type": "application/json"
            },
            json={
                "agent_id": os.environ["TALKIFAI_AGENT_ID"],
                "user_identifier": email,
                "metadata": metadata or {}
            }
        )
        
        if response.status_code != 200:
            raise HTTPException(
                status_code=response.status_code,
                detail=response.json()
            )
        
        return response.json()

React (Frontend with Backend Proxy)

// ChatWidget.tsx
async function startChat(userEmail: string) {
  // Call YOUR backend, not TalkifAI directly
  const response = await fetch('/api/chat/start', {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      email: userEmail,
      metadata: {
        source: 'chat_widget',
        page: window.location.pathname
      }
    })
  });

  if (!response.ok) {
    const error = await response.json();
    throw new Error(error.detail || 'Failed to start chat');
  }

  const data = await response.json();
  
  // Store session token for subsequent requests
  sessionStorage.setItem('chat_session_token', data.session_token);
  sessionStorage.setItem('conversation_id', data.conversation_id);
  
  // Show greeting if agent speaks first
  if (data.greeting) {
    addMessage('assistant', data.greeting);
  }
  
  return data;
}

Best Practices

1. Always Use Backend Proxy

Never expose API key in frontend: ❌ Wrong:

// DON'T DO THIS - exposes API key
fetch('https://api.talkifai.dev/v1/chat/sessions', {
  headers: {
    'X-API-Key': 'tk_live_...' // Exposed in browser!
  }
});

✅ Correct:

// Call your backend instead
fetch('/api/chat/start', {
  method: 'POST',
  body: JSON.stringify({ email: user@email })
});

2. Provide User Identifier

Always provide user_identifier for returning user memory:

{
  "user_identifier": "customer@example.com" // ✅ Good
}

Without it, every session is isolated and memory cannot be retrieved.

3. Handle Greeting Properly

Check if greeting exists before displaying:

if (data.greeting) {
  addMessage('assistant', data.greeting);
}
// If null, agent waits for user to speak first

4. Store Session Token Securely

Browser:

sessionStorage.setItem('chat_session_token', data.session_token);

React Native:

await AsyncStorage.setItem('chat_session_token', data.session_token);

5. Handle Errors Gracefully

try {
  const response = await startChat(email);
} catch (error) {
  if (error.status === 402) {
    showCreditError();
  } else if (error.status === 401) {
    showAuthError();
  } else {
    showGenericError();
  }
}

Send Message — Send messages with SSE streaming
Get History — Retrieve message history
End Session — End session and cleanup
Create Voice Session — LiveKit voice session

Next Steps

Send Message

Use the session_token to send messages via SSE streaming.

Integration Guide

See complete examples in the Chat API Guide.

Authorizations

x-api-key

string

header

required

Your TalkifAI API key. Generate from Studio → Settings → API Keys. Format: tk_live_...

Body

application/json

agent_id

string

required

ID of a Text architecture agent

Example:

"5b710eca-ee67-4c3a-aeb6-8b541f451b40"

Response

Session created

conversation_id

string

Example:

"chat_5b710eca_user123_1705312800000"

session_token

string

JWT valid for 24 hours

agent_name

string

Example:

"Support Agent"

greeting

string | null

Example:

"Hello! How can I help you today?"

agent_model

string

Example:

"gpt-4o-mini"

Chat API Reference Send Message (SSE)

​Overview

​Request

​Request Fields

​User Identifier (Important)

​Response

​Response Fields

​Authentication Modes

​Mode 1: API Key (External Embeds)

​Mode 2: Studio Token (Internal Preview)

​What Happens on Session Creation

​Error Responses

​400 Bad Request

​401 Unauthorized

​402 Payment Required

​404 Not Found

​503 Service Unavailable

​Examples

​cURL (API Key)

​JavaScript (Backend Proxy)

​Next.js API Route

​Python (FastAPI)

​React (Frontend with Backend Proxy)

​Best Practices

​1. Always Use Backend Proxy

​2. Provide User Identifier

​3. Handle Greeting Properly

​4. Store Session Token Securely

​5. Handle Errors Gracefully

​Related Endpoints

​Next Steps

Send Message

Integration Guide

Authorizations

Body

Response

Overview

Request

Request Fields

User Identifier (Important)

Response

Response Fields

Authentication Modes

Mode 1: API Key (External Embeds)

Mode 2: Studio Token (Internal Preview)

What Happens on Session Creation

Error Responses

400 Bad Request

401 Unauthorized

402 Payment Required

404 Not Found

503 Service Unavailable

Examples

cURL (API Key)

JavaScript (Backend Proxy)

Next.js API Route

Python (FastAPI)

React (Frontend with Backend Proxy)

Best Practices

1. Always Use Backend Proxy

2. Provide User Identifier

3. Handle Greeting Properly

4. Store Session Token Securely

5. Handle Errors Gracefully

Related Endpoints

Next Steps