Skip to content

Quickstart

Get Remembra running in 5 minutes.

Prerequisites

  • Docker (recommended) or Python 3.10+
  • OpenAI API key (for embeddings/extraction)

Step 1: Start Remembra

docker run -d \
  -p 8787:8787 \
  -e OPENAI_API_KEY=sk-your-key \
  -v remembra-data:/app/data \
  remembra/remembra

Create docker-compose.yml:

version: '3.8'
services:
  remembra:
    image: remembra/remembra
    ports:
      - "8787:8787"
    environment:
      - OPENAI_API_KEY=${OPENAI_API_KEY}
    volumes:
      - remembra-data:/app/data

volumes:
  remembra-data:

Then run:

docker-compose up -d

git clone https://github.com/remembra/remembra
cd remembra
pip install -e ".[server]"

export OPENAI_API_KEY=sk-your-key
python -m remembra.server

Step 2: Verify It's Running

curl http://localhost:8787/health
# {"status":"healthy","version":"0.6.3"}

Or open the dashboard: http://localhost:8787

Step 3: Install the SDK

pip install remembra

Step 4: Store Your First Memory

from remembra import Memory

# Connect to your Remembra instance
memory = Memory(
    base_url="http://localhost:8787",
    user_id="quickstart-user"
)

# Store a memory
memory.store("""
    Had a great meeting with Sarah from Acme Corp today.
    She mentioned they're looking for AI solutions for their
    customer support team. Budget is around $50k/year.
    Follow up next Tuesday.
""")

print("Memory stored!")

Step 5: Recall Memories

# Ask questions about your memories
context = memory.recall("What do I know about Acme Corp?")
print(context)
# Output: "Sarah from Acme Corp is looking for AI solutions 
#          for customer support. Budget: $50k/year. 
#          Follow up scheduled for Tuesday."

What Just Happened?

  1. Smart Extraction: Your messy text was transformed into clean facts
  2. Entity Resolution: "Sarah" was identified as a PERSON, "Acme Corp" as an ORG
  3. Relationship Mapping: Sarah → WORKS_AT → Acme Corp
  4. Vector Storage: Facts embedded and stored for semantic search
  5. Recall: Your query found the relevant memories

Next Steps

Example: Building a Chatbot

from remembra import Memory
import openai

memory = Memory(base_url="http://localhost:8787", user_id="user_123")

def chat(user_message: str) -> str:
    # Recall relevant context
    context = memory.recall(user_message, limit=5)

    # Build prompt with memory
    messages = [
        {"role": "system", "content": f"You are a helpful assistant. Context: {context}"},
        {"role": "user", "content": user_message}
    ]

    # Get response
    response = openai.chat.completions.create(
        model="gpt-4o",
        messages=messages
    )

    assistant_message = response.choices[0].message.content

    # Store the conversation
    memory.store(f"User: {user_message}\nAssistant: {assistant_message}")

    return assistant_message

# Chat with memory!
print(chat("My name is Alex and I love hiking"))
print(chat("What do you know about me?"))  # Remembers Alex loves hiking!

Pro Tip

Store important facts explicitly, not just conversation history. The extraction model works best with clear statements.