Investors
The versioning layer for MongoDB, built for the agent era
Postgres has Neon. MySQL has PlanetScale. MongoDB — the default database of a generation of application developers — has no equivalent. Argon fills that gap, at the moment AI agents make disposable, auditable, undoable databases a necessity.
Where it stands
- Open-source engine (MIT), v2.1.3 shipped: branching, time travel, data PRs, per-actor undo
- Agent surface live: MCP server, TTL sandboxes, dataset pins, Python SDK 0.2.0 available from its Git release
- Public benchmark suite separates metadata costs from checkout, capture and end-to-end workflows
- Real-driver compatibility validated in CI (pymongo, mongoose)
Where it goes
- Model: open-source core, managed cloud on top — the path Neon validated
- Market: MongoDB users needing versioning; AI/ML teams needing sandboxes and reproducible evals
- Wedge: the undo button for AI agents — a safety primitive every agent deployment needs
Seeking strategic investors to scale the engine into a managed service. For the deck and a conversation: