Quick Start

Get your first working integration with the Memory Platform in less than 5 minutes.

1. Prerequisites

• Python 3.9+ or Node.js 18+
• Memory Platform API Key (Contact your administrator)
• Tenant ID (The namespace for your data)

2. Install the SDK

Install the standard Python SDK to interact with the Memory Engine:

pip install memory-platform

3. Store Your First Memory

Create a memory block about a specific subject (e.g., a User).

from memory_platform import MemoryClient

client = MemoryClient(api_key="your-api-key")

# Store a preference for a user
client.memory.upsert(
    tenant_id="acme-corp",
    subject={
        "type": "user",
        "id": "user_99",
        "display_name": "Alice Smith"
    },
    kind="preference",
    content="User prefers high-level summaries over technical details"
)

4. Retrieve Relevant Knowledge

Use semantic search to find memories related to a query, scoped to a specific subject.

# Search memories for the same user
memories = client.memory.search(
    query="How should I present this report?",
    tenant_id="acme-corp",
    dimension_scope={
        "subject_type": "user",
        "subject_id": "user_99"
    }
)

print(memories[0].content.text) 
# Output: "User prefers high-level summaries over technical details"

5. What's Next?

• Core Concepts: Learn about Salience, Stability, and the Knowledge Graph.
• Document Intelligence: Learn how to link memories back to PDF source coordinates.
• MCP Integration: Give your Claude or GPT agent native access to this memory.