Skip to main content

lawmem.ai Developer Documentation

lawmem.ai is a Memory-as-a-Service platform for legal AI agents. It gives your agents persistent semantic memory — store context, recall it by meaning, and maintain continuity across sessions.

The Problem​

Most AI agents are stateless. Every session starts from zero. For legal AI agents, this creates real problems:

  • A contract review agent forgets every prior negotiation
  • A legal research agent cannot build on prior session context
  • A client intake agent has no memory of previous interactions

lawmem.ai solves this with a simple API: store text, recall it by semantic similarity, delete when done.

How It Works​

  1. Your agent calls POST /store with a text payload and namespace
  2. lawmem.ai embeds the text using nomic-embed-text (768-dim vectors)
  3. Vectors are stored in Qdrant with PostgreSQL metadata, isolated by namespace
  4. Your agent calls POST /recall with a query and gets results ranked by cosine similarity
  5. Each call is metered per your plan

Key Features​

  • MCP-native — plug into any MCP-compatible agent framework
  • Namespace isolation — each client or matter gets its own isolated memory space, enforcing attorney-client privilege boundaries at the infrastructure level
  • Pay-per-call or subscribe — USDC on Base via x402, or monthly subscription
  • LUKS2 encrypted at rest — AES-256, argon2id PBKDF
  • GDPR Article 28 compliant — DPA available on Legal Pro plan