Local semantic search
Index project folders, runbooks, Markdown files, decision logs, and notes into a searchable local memory store.
Private local memory
Agent Memory
A local semantic memory layer for AI agents, assistants, and project teams that need useful context without sending private files to a hosted vector database.
API ready at http://127.0.0.1:8765
Memory setup
Add knowledge sources
Create your user guidance profile.
Choose folders or files to index.
Search memory or connect an agent.
Memory search
Ask Agent Memory
Agent Guidance
About Me
How I Like to Work
Very Important
Built for continuity
Give every assistant the context it should have had already.
Agent guidance
Save “About Me,” working preferences, and non-negotiables so assistants receive consistent context before search results.
Local API
Connect compatible agents and automations to a local HTTP API running on your machine.
Installable workflow
Designed as a lightweight Windows app that uses Edge app mode and keeps the desktop package small.
Privacy model
Local-first by default.
Agent Memory stores indexed content on the user’s machine and binds the local API to 127.0.0.1. The default MiniLM embedding model is downloaded on first use and then cached locally.
Localmachine storage
MiniLMsemantic embeddings
127.0.0.1local API binding
Windows package
Clear install path, commercial terms up front.
The installer starts with a readable legal agreement and requires explicit acceptance before the local service is installed or started.
- 1
Download and extract the Windows package.
- 2
Open Double Click to Install.
- 3
Review and accept the Commercial License, Terms of Use, and Attributions.
- 4
Start indexing local knowledge sources.
Commercial use
Proprietary product. Permissive components.
Agent Memory is proprietary commercial software from so.design. Current dependency and model audits found no GPL, AGPL, LGPL, or unknown-license runtime blockers. Third-party components remain governed by their own licenses.