Connecting the Dots: A Secure Protocol for Cross-Team Knowledge DiscoveryDraft

Large organizations solve the same problems in parallel without knowing it. Two teams build similar features, three groups research the same vendor, and nobody realizes until the work is done. The information exists to connect these efforts — but surfacing it requires access to everyone's work, which creates privacy, political, and security problems that no organization will accept.

The Semantic Vault is a protocol that solves this paradox: it discovers semantic relationships across teams without exposing anyone's work. Using encrypted embeddings, trusted execution environments, and a bilateral consent protocol, it can identify that two teams are working on related problems and facilitate an introduction — without ever revealing what either team is actually doing until both explicitly opt in.

This paper specifies the complete protocol across registration, matching, and consent phases, and evaluates both centralized and federated deployment architectures.

What You'll Learn

• Why cross-team knowledge discovery is harder than it looks: the privacy, trust, and political barriers that kill naive approaches
• The Semantic Vault protocol: registration, matching, and bilateral consent in detail
• How trusted execution environments (TEEs) enable computation on encrypted data without trusting the platform operator
• Centralized versus federated deployment architectures, with trade-offs for each
• Integration patterns with existing enterprise systems including Slack, Confluence, and internal knowledge bases

Who This Is For: Engineering directors, knowledge management leads, and platform architects in organizations large enough that teams regularly duplicate effort.