Introduction to Liebed

Liebed is a developer-first memory layer that gives AI agents persistent, intelligent long-term memory. Instead of building custom RAG pipelines or struggling with context limits, Liebed lets your agents automatically extract, store, and recall relevant facts over time.

Why Liebed?

Traditional vector databases require you to handle embedding, chunking, and semantic search manually. More importantly, they only retrieve text snippets. Liebed actually understands the memories:

• Automatic Extraction: Send raw conversation logs; we extract the facts.
• Contradiction Handling: If a user moves from "New York" to "London", Liebed updates the truth state automatically.
• Temporal Awareness: Liebed knows the difference between what happened yesterday and what happened a year ago.

Quick Example

Initialize the client and add a memory.

import { Liebed } from '@liebed/sdk';

const liebed = new Liebed({
  apiKey: process.env.LIEBED_API_KEY
});

// Save a memory
await liebed.remember({
  user_id: "user_123",
  content: "User prefers concise answers and is building an AI startup."
});