The JuLenny Vault
Engineering insights on Fully Homomorphic Encryption, privacy-preserving computation, and building on encrypted data.
JuLenny is now available on Google Cloud Marketplace
You can now buy JuLenny through Google Cloud Marketplace and pay for it on your existing Google Cloud account. Same platform, same plans, one less vendor to set up.
David Uzan · 24 Aug 2026
JuLenny is live: let a partner compute on your data without ever exposing it
JuLenny launches its multi-party data collaboration platform. Organizations run agreed computations on combined data and nobody, not even JuLenny, sees the inputs. Five ready functions, a source-available client, and an MCP server for AI agents.
David Uzan · 18 Aug 2026
Checking for conflicts when neither side can show their data
Drug interactions, allergens, compliance violations: the rules are public, the data is private. Rule-Based Cross-Match checks one party's list against another's under shared rules, and only the matches surface.
David Uzan · 17 Aug 2026
Every rejected counter-offer tells your counterparty something. It doesn't have to.
Negotiation leaks: every refusal maps your constraints a little more. The Encrypted Negotiation Matrix tells two parties whether a deal exists without revealing either side's position, and failed rounds leak nothing at all.
David Uzan · 14 Aug 2026
Run your model on their data. See neither.
A bank's credit model and a fintech's customer data would profit from meeting, but neither side will move first. Decision Tree Inference runs one party's model on another party's encrypted data, and only the classification comes out.
David Uzan · 14 Aug 2026
Two companies, one AI model, zero shared data
Organizations everywhere train the same ML models separately, and every one of them is worse than the model they could build together. Federated Average combines locally trained models without exposing training data or weights to anyone.
David Uzan · 10 Aug 2026
How many records do we share? Now you can answer without showing your data.
Two organizations can now count their shared records, customers, threat indicators, SKUs, without either side revealing its list to anyone. Here is how Joint Record Overlap works and where it applies, from co-marketing to post-cookie ad attribution.
David Uzan · 31 Jul 2026