Developers

Build with Rabbit

Add memory intelligence to any application. Rabbit's API is OpenAI-compatible, so if you've used a chat completion endpoint before, you already know how to use this. Extract entities, classify intent, detect relationships, and get cited answers from organizational data.

Quickstart API Reference Try Sandbox

Quickstart

Three steps to your first extraction

Get your API key

# Set your API key
export RABBIT_API_KEY="rb_live_sk_..."

Extract from any text

Send raw text to the extract endpoint. Get back structured entities, decisions, and action items as JSON.

curl https://rabbit.reattend.com/v1/extract \
  -H "Authorization: Bearer $RABBIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Met with Sarah from Acme.
    Budget confirmed at $45,000."
  }'

Get structured output

Every response returns clean JSON with extracted entities, decisions, action items, dates, and topics. Ready for your database.

{
  "people": ["Sarah"],
  "organizations": ["Acme"],
  "decisions": [
    "Budget confirmed at $45,000"
  ],
  "action_items": [{
    "owner": "Sarah",
    "task": "Send contract"
  }]
}

API Reference

Endpoints

All endpoints accept JSON and return JSON. Authentication via Bearer token. Base URL: https://rabbit.reattend.com

Extract entities and facts

Pulls people, organizations, decisions, action items, dates, and topics from unstructured text. Works with meeting notes, emails, Slack messages, and any organizational content.

Parameters

Name	Type	Required	Description
text	string	required	The text to extract from
signals	string[]	optional	Specific signals: people, decisions, actions, dates, topics

Response

{
  "people": ["Sarah Chen"],
  "organizations": ["Acme Corp"],
  "decisions": ["Budget confirmed at $45,000"],
  "action_items": [{
    "owner": "Sarah Chen",
    "task": "Send signed contract",
    "due": "Friday"
  }],
  "dates": ["Tuesday", "Friday"],
  "topics": ["Q2 launch", "contract"],
  "latency_ms": 1847
}

Classify and summarize

Auto-triages incoming content into a type (meeting, decision, task, insight, note) with a summary, tags, importance score, and sentiment.

Parameters

Name	Type	Required	Description
text	string	required	The content to triage

Response

{
  "type": "meeting",
  "summary": "Meeting with Sarah Chen from Acme Corp...",
  "tags": ["acme-corp", "q2-launch", "budget"],
  "importance": 0.88,
  "sentiment": "positive"
}

Ask questions, get cited answers

Given a question and a set of memory contexts, generates a conversational answer with numbered source citations, confidence score, and follow-up suggestions.

Parameters

Name	Type	Required	Description
question	string	required	Natural language question
context	object[]	required	Array of memory objects to search
stream	boolean	optional	Stream the response (SSE)

Response

{
  "answer": "The Q2 launch was confirmed for April 22 [1]...",
  "sources": [1],
  "confidence": 0.96,
  "follow_ups": [
    "Has Sarah sent the contract?",
    "What is the compliance review status?"
  ]
}

Classify query intent

Classifies a natural language question into one of 7 intent types to route to the right retrieval strategy.

Parameters

Name	Type	Required	Description
question	string	required	The user's question

Response

{
  "intent": "synthesis",
  "confidence": 0.97,
  "latency_ms": 270
}

Detect relationships

Given two pieces of content, detects the relationship type: follows_up, contradicts, supports, expands, references, supersedes, or relates_to.

Parameters

Name	Type	Required	Description
source	string	required	First memory content
target	string	required	Second memory content

Response

{
  "relationship": "follows_up",
  "confidence": 0.89,
  "explanation": "The second memory references the budget decision from the first."
}

Analyze sentiment

Detects the emotional tone of organizational communications: positive, negative, neutral, mixed, or urgent.

Parameters

Name	Type	Required	Description
text	string	required	The text to analyze

Response

{
  "sentiment": "positive",
  "confidence": 0.91,
  "latency_ms": 350
}

Sandbox

Try it now

No API key needed. Paste any text and see Rabbit process it.

Input

Output

// Click "Run Rabbit" to process

Rate Limits & Pricing

Start free. Scale as you grow.

Beta developers get free access. Enterprise plans have no rate limits.

Plan	Rate limit	Signals	Support	Price
Beta (current)	100 requests/day	All 12 signals	Community	Free
Developer	10,000 requests/day	All 12 signals	Email	Coming soon
Team	100,000 requests/day	All 12 signals + custom	Priority	Coming soon
Enterprise	Unlimited	All + custom fine-tuning	Dedicated engineer	Contact us

SDKs & Libraries

Use Rabbit in your language

OpenAI-compatible API means you can use any existing OpenAI SDK today. Native SDKs coming soon.

Python

pip install rabbit-ai

Coming soon

Node.js

npm install @rabbit/sdk

Coming soon

curl

REST API

Works today via curl

Available

oai

OpenAI SDK

Compatible endpoint

Available

Start building with Rabbit

Get your API key and make your first extraction in under 5 minutes.

Get API key Talk to us