How It Works
1
Set up your environment
Before using the core APIs, create the required resources:
- Create a Workspace — the namespace that isolates your data
- Create an Actor — the identity representing your end user
- Create a Project — the container for documents, conversations, and facts
2
Feed memories
Import knowledge into a project through two channels:
- Documents — upload files to the Library, then import them into a project. MemoryLake indexes the content for semantic search.
- Conversations — create a conversation, then append messages. MemoryLake automatically extracts structured facts from the exchange.
3
Query memories
Search across all memory sources in a workspace using natural language. Results include matched document passages and extracted facts.
Core APIs
Import Documents
Import files from the Library into a project for indexing.
Create Conversation
Start a new conversation in a project to capture interactions.
Append Message
Add a message to a conversation. Facts are extracted automatically.
Search Memories
Query documents and facts across a workspace with natural language.