engramBring your knowledge in.
Import documents and ChatGPT exports. Capture a browser page, selection or note. Keep the provenance that makes a fact useful.
Explore capture →PERSONAL MEMORY, ON YOUR TERMS
Your knowledge.
Your context.
Your server.
Stop explaining everything from scratch. Bring your documents, saved pages and conversations into one memory layer for you and your AI clients.
Free for personal, noncommercial use. Commercial rights reserved.
LOCAL OLLAMAPOSTGRESQL + PGVECTORHUMAN REVIEWMCP + RESTENCRYPTED BACKUPS
FROM SCATTERED SOURCES TO SHARED CONTEXT
A memory you can
actually work with.
You decide what sticks.
Review extracted candidates in the Memory Inbox. Inspect proposed updates and conflicts before accepting changes to your knowledge.
See projects and review →Put context to work.
Search with semantic and full-text retrieval. Keep projects separate and connect AI clients through scoped MCP access or the REST API.
Connect an MCP client →DESIGNED TO BE RECOVERED
Your knowledge should
outlive your server.
Encrypted backups are part of Engram's architecture. Isolated restore tests check the database and the document objects together.
Read the recovery evidence →DOCUMENTED RECOVERY EXERCISE · 14 SEP 2026
- Memories restored
- 672
- Documents restored
- 422
- Missing object references
- 0
One recorded off-site exercise, completed in 17 seconds. Recovery time depends on data, hardware and configuration.
BUILD YOUR OWN CONTEXT LAYER
Make room for
what you know.
For technically comfortable personal users. Bring a Linux host, Docker Compose and room for your knowledge.
What do I need to run Engram?
The deployment guide recommends 4 CPU cores and 8 GB RAM, with 16 GB or more preferred. Allow storage for PostgreSQL, documents, Ollama models and encrypted backups. Configure secrets and an administrator before use.
Is Engram open source?
Engram is source-available and free for personal, noncommercial self-hosting. Workplace use, commercial integrations and managed hosting require separate written permission. Read the license and usage guide.
Where does AI processing happen?
Engram uses Ollama for local embeddings and extraction on your infrastructure. Optional connectors communicate with their external services. You control which sources and clients you connect.
Can I use the mobile app or buy managed hosting?
The stable server release is v2.7.0. Native iOS and Android companions remain preview work with device qualification still open; Gmail is also a preview connector. This page offers self-hosting, not a managed subscription. For commercial enquiries, contact Jayden.