Supermemory is a memory API and knowledge layer that turns anything you read, save or say into a searchable second brain for you and your AI agents.
Supermemory captures content from the web, PDFs, notes and apps, builds a structured knowledge graph around it, and makes it instantly retrievable through an API or browser extension. When an agent asks a question, Supermemory returns the relevant context with citations instead of raw similarity matches.
Designed as infrastructure, Supermemory powers personalized assistants, research tools and agents that need a user’s real context. It handles ingestion pipelines, chunking, deduplication and incremental updates, and its graph-backed retrieval keeps answers grounded in what the user actually saved. Developers can scope memories per user, per project or per agent with a single parameter.
Supermemory offers a free tier for personal use with a storage and query allowance. Paid plans scale by stored items and API calls, and enterprise tiers add dedicated infrastructure, custom retention and support. Pricing stays usage-based so early products are not penalized. Open-source self-hosting is available for teams that must keep data inside their own perimeter.
Supermemory is for developers building context-aware products and for power users who want a personal knowledge layer their tools can share. It fits teams building research assistants, reading companions, customer copilots and agent products that need durable user context.
Yes, individuals can use Supermemory for free within the starter limits, including the browser extension and personal search. Developers shipping to customers move to paid usage-based tiers as their memory stores and query volume grow.
Yes, the memory API is the core of the product. It exposes endpoints for adding content, searching memories and injecting context into prompts, with SDKs and documentation aimed at production agent developers.
Supermemory goes beyond note-taking apps by treating memory as infrastructure: content is structured, cited and served programmatically to agents. Compared with plain vector databases it handles ingestion, graph relationships and personalization out of the box. Bookmarking tools organize what you saved; Supermemory makes that knowledge machine-usable by the agents you run.