# Integrating AI Chat

AI Chat is the simplest and fastest way to integrate therappai’s AI therapist into your product.\
Your application sends a user message to the API, and the API responds with supportive, therapy-aligned guidance. This flow powers chat interfaces, onboarding conversations, coaching tools, and lightweight support experiences.

This page shows how to implement AI chat end-to-end.

***

### **How AI Chat Works**

Each chat interaction follows a simple loop:

1. User sends a message
2. Your app sends that message to `/chatting/`
3. therappai generates a therapist response
4. Your app displays the response in your UI
5. Repeat for the conversation

There’s no session object to create or maintain — the API is stateless per request.

***

### **1. Collect the User’s Message**

Your UI (mobile, web, chat bubble) captures the text the user wants to send.\
Example:

```
"I'm feeling overwhelmed today. I'm not sure why."
```

You then package this into a JSON request.

***

### **2. Send Message to therappai**

Make a POST request to:

```
POST /chatting/
Authorization: Bearer ACCESS_TOKEN
Content-Type: application/json
```

Body example:

```json
{
  "message": "I'm feeling overwhelmed today and can't focus."
}
```

Your app can include additional context if needed (optional).

***

### **3. Receive the AI Therapist Response**

The API returns a structured response that includes the therapist’s message.

Example:

```json
{
  "reply": "I'm really sorry you're feeling overwhelmed. Let’s take one step at a time. Can you tell me what part of today has been the hardest so far?"
}
```

Your UI displays this however you choose:

* chat bubble
* message card
* conversation feed
* coaching panel

This keeps your UX completely flexible.

***

### **4. Render the Response in Your UI**

Most developers show AI chat using:

* a chat bubble layout
* timestamped messages
* typing indicator (“AI is responding…”)
* scrollable conversation feed
* optional “suggested replies”

The message structure is intentionally simple so you can style it however you want.

***

### **5. Repeat the Loop**

Each time the user sends a message, you:

1. Call `/chatting/`
2. Display the therapist’s response
3. Store the chat if your UI needs history

This continues until the interaction ends.

***

### **Optional: Store Chat History**

Because the API is stateless, you decide:

* whether to store conversations
* how long to store them
* whether to show history in your app
* whether to sync chats across devices

Many early-stage integrations simply store messages client-side.

***

### **Recommended UX Enhancements**

Developers often add:

#### **Typing Indicators**

Show a spinner or “therappai is responding…” after sending a message.

#### **Message Error Handling**

If network fails or token expires, show a retry option.

#### **Suggested Replies**

Prompt users with one-tap responses generated by your own heuristics or content.

#### **Protective Copy**

For sensitive messages, include calm, grounding microcopy.

***

### **Safety Layer**

Every message is processed through therappai’s safety system to identify:

* risk-related language
* self-harm signals
* severe distress phrases
* harmful or abusive content

The API always responds with a **safe, supportive message**.\
No raw model output is returned.

> **Note:** therappai does NOT contact emergency services.\
> Crisis Buddy contacts can be shown by *your* app if appropriate.

***

### **Minimum Viable Integration (5 Steps)**

1. Login user → get access token
2. Build a text input field
3. Send messages to `/chatting/`
4. Display therapist responses
5. Refresh token when needed

That’s enough to launch a fully working AI therapy chat.