Sources
How AI chats, social history, browser activity, and files can enter a local DeepMem vault over time.
Documentation
DeepMem Docs
Understand the DeepMem model: sources, signals, recall, agent grants, and privacy controls for local-first memory.
Product model
Sources -> Signals -> Recall -> Agent Grants
DeepMem starts by bringing source history back to this computer, then turns it into reviewable signals. Recall lets the user ask memory with evidence, and Agent Grants control which selected context can be used by tools.
Signals
How raw traces become reviewable facts, preferences, decisions, procedures, insights, and next steps.
Recall
How to ask your past and inspect the evidence bundle before copying or sharing context.
Agent Grants
How future local tools and agents request scoped memory instead of receiving an archive.
Privacy
What local-first means, what requires permission, and which controls should remain visible.
Privacy model
Local-first by default, evidence-backed by design.
DeepMem is being designed so captured records, indexes, and derived signals stay inspectable. Agent access should be scoped by user permission rather than all-or-nothing.
Is DeepMem available now?
DeepMem is in active development. Join the waitlist for early access updates.
Does DeepMem upload my data?
The product direction is local-first by default. Any future remote, sync, or cloud AI feature should be explicit and permissioned.
Will all listed connectors ship at launch?
No. The site describes the product direction. Connectors will roll out gradually and stay labeled as early access, planned, or research.
Can DeepMem fully import WeChat history on macOS?
Full local WeChat database history may require advanced local permissions. DeepMem will not make disabling SIP part of the default user path.
Waitlist
Join the DeepMem waitlist
Get early access updates for the desktop app and help shape the signals, recall, and agent-grant workflow.