Manage · Self-Hosted
In developmentOrganizational memory that runs entirely on your own hardware. The same source-backed, human-reviewed knowledge engine behind Revery, for teams whose data can't live in someone else's cloud.
Get NotifiedInvention teams protecting unfiled IP. Regulated firms with client data rules. Companies that have simply decided their institutional memory is too valuable to rent back from a vendor. For them, the company-brain question comes with a constraint the hosted tools can't meet: the knowledge, and the AI reading it, have to stay on hardware they control.
Recall is built for that constraint. A single self-hosted server that ingests your sources (interviews, meetings, documents, decisions) and maintains evidence-backed memory: every claim tied to its sources, every promotion to "trusted" made by a human, changes and conflicts tracked over time.
It works with local AI models (an air-gapped deployment makes no outside calls at all), your team's agents connect over MCP with read-and-propose access only, and each collection lives in a single portable file you can back up, move, or hand over.
Same method, same trust model, different home. Revery is hosted and ready for teams that want the benefits without running anything. Recall is for teams where self-hosting is the requirement that comes first: your hardware, your models, your rules.
If you're not sure which side of that line your work falls on, that question (which tasks can use the cloud and which can't) is one I help teams answer all the time. A 15-minute call usually settles it.
It's the engine I'm building the whole method on, and the self-hosted edition ships to a small group of early teams first. Leave your email and I'll let you know when it's ready.
Get Notified