AI agents are getting smarter, but they still struggle to remember what matters across sessions, tools and applications. Lighthouse Memory is building a shared, persistent memory layer for AI agents. It allows connected agents like Claude, Codex, Cursor and other AI applications to save, retrieve and reuse important context, including user preferences, project history and past decisions. Instead of starting from scratch every time, agents can pick up where the previous session or connected agent left off.
AI agents can write code, conduct research and automate complex tasks, but most still operate with fragmented memory. When users switch tools, start new sessions or move between agents, important context often gets lost.
Lighthouse Memory solves this by giving AI agents a persistent memory layer that exists independently of the models and applications using it.
Lighthouse allows connected AI agents to remember useful information, retrieve relevant context and continue working across sessions.
For example, a developer working with Claude Code can save project requirements, architecture decisions and coding preferences. When they switch to Codex or another connected agent, that agent can retrieve the saved context without requiring the developer to explain everything again.
Shared memory across agents: Connected AI tools can access a common memory backend.
Persistent context: Preserve project knowledge, user preferences and past decisions across sessions.
Semantic recall: Retrieve relevant memories based on meaning rather than exact keywords.
MCP and SDK integration: Developers can add memory to supported AI clients and custom agents.
Verifiable memory: Content-addressed records allow applications to verify the integrity of stored information.
Flexible storage: Support for IPFS, Filecoin, Walrus and user-controlled storage such as S3, R2, Supabase and MinIO.
Privacy options: SEAL-encrypted memory through the Memwal engine on Walrus.
As AI agents become part of everyday workflows, their ability to retain and share context becomes essential.
Lighthouse is building the memory infrastructure that allows agents to work with continuity, whether they are assisting developers, conducting research, managing workflows or collaborating with other agents.
The goal is simple: users should be able to change AI tools without losing the context they've already built.
Your agents remember. Everywhere