Memori + Gradient AI

Memori + DO Gradient

Tell me your name, then refresh and ask "What's my name?"

Extracted memories

No memories extracted yet.
Start chatting to see extracted facts!

Test with cURL

Send your first request to see memory in action:

curl -X POST http://generalmillsnacufs2020.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4",
    "messages": [{"role": "user", "content": "My name is Alex"}],
    "user": "user-123"
  }'

How Memory Works

1. Send messages with a user ID
2. Memori automatically extracts facts from conversations
3. On future requests, relevant memories are injected into context
4. The AI remembers everything about each user

Tip: The user field is required. Use your app's user ID so each user has their own memory.

Python (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="http://generalmillsnacufs2020.com/v1/",
    api_key="not-needed"
)

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "I love hiking"}],
    user="user-123"  # Required for memory
)
print(response.choices[0].message.content)

JavaScript / Node.js

import OpenAI from 'openai';

const client = new OpenAI({
    baseURL: 'http://generalmillsnacufs2020.com/v1/',
    apiKey: 'not-needed'
});

const response = await client.chat.completions.create({
    model: 'gpt-4',
    messages: [{ role: 'user', content: 'I work at Acme Corp' }],
    user: 'user-123'
});
console.log(response.choices[0].message.content);

Drop-in replacement: Any OpenAI-compatible SDK works. Just change the base URL.

Endpoints

Method	Endpoint	Description
POST	`/v1/chat/completions`	Chat with memory
GET	`/v1/memories/{user_id}`	Get user memories
DELETE	`/v1/memories/{user_id}`	Clear user memories
GET	`/v1/usage`	Memory usage stats
GET	`/health`	Health check

Chat Request Body

{
  "model": "gpt-4",           // Any model name (routed to Gradient AI)
  "messages": [...],          // OpenAI message format
  "user": "user-123",         // Required: unique user identifier
  "temperature": 0.7,         // Optional: 0-2
  "max_tokens": 1000          // Optional
}

Response Extensions

Responses include a memori field with memory stats:

{
  "choices": [...],
  "memori": {
    "facts_extracted": 2,     // New facts from this message
    "facts_recalled": 3,      // Memories used for context
    "memories_used": 150,     // Total memories stored
    "memories_limit": 5000    // Your quota
  }
}

Option 1: Call the API (Recommended)

Use this endpoint from anywhere - your server, frontend, or serverless functions:

http://generalmillsnacufs2020.com/v1/chat/completions

Works with any language or framework that supports HTTP requests or OpenAI SDKs.

Option 2: Build on the Droplet

Deploy your own app alongside Memori on the same server:

# SSH into your Droplet
ssh root@your-droplet-ip

# Call the local API from your app
curl http://localhost:8000/v1/chat/completions ...

For Droplet management, see DigitalOcean Droplet Docs.

Need More Memories?

This deployment includes 5,000 free memories. For unlimited usage:

Memori Cloud - Managed, unlimited memories
Self-host Memori - Full control, Apache 2.0 license

Resources: Memori Docs | Gradient AI Docs | Discord