Memory you can commit to.Git-first context for AI agents.
vem turns agent context into durable project memory stored in Git. Your team gets verifiable tasks, decisions, and snapshots with cloud indexing for lightning-fast retrieval.
update_task({ id: "TASK-003", status: "done" })Vector search + analytics, no data lock-in.
Memory Loop
From prompt to shared memory in minutes.
Capture the right context, commit it to Git, and let the indexer make it searchable across every agent and teammate.
Capture
Pack conversations into durable tasks, decisions, and context.
Commit
Every update is a Git commit with evidence and history.
Retrieve
Cloud indexing powers fast semantic search without owning data.
Why vem
Memory you can prove, not just store.
vem is a project memory system, not another vector DB. It aligns agents and developers around a single, verifiable source of truth so every context update is inspectable, shareable, and trusted.
Git is the source of truth
All memory artifacts live in your repo, so history and ownership stay with you.
Agents are transient, context is durable
The .vem/ folder persists across machines, agents, and handoffs.
Cloud indexes, Git owns
The SaaS layer accelerates search and analytics without holding your data.
What ships with vem
Git-native memory
Snapshots, tasks, and decisions are versioned alongside code.
Agent + human parity
MCP tools and the CLI read/write the same memory state.
Context orchestration
Pack, prune, and sync context with evidence for every update.
Cloud-indexed search
Semantic search and analytics without losing ownership of data.
Secure by default
Redaction, tenant isolation, and audit-friendly memory trails.
Instant retrieval
Find the right memory fast, even across large histories.
Start building
Ready to upgrade your agents?
Give every agent a shared memory layer with Git-backed history, searchable context, and verifiable decisions.