Anna Rose sits down with Gnosis Guild founder Auryn Macmillan for a technically dense walkthrough of two interconnected infrastructure projects. The first half covers Zodiac's evolution: what began as a DAO toolset for managing treasuries and multisigs has matured into a modular access control primitive that any protocol can plug into. Auryn explains the specific architectural decisions that enabled this shift — separating policy from mechanism, standardizing module interfaces, and leaning into composability over monolithic design.
The second half dives into The Interfold, an encrypted execution layer that combines fully homomorphic encryption, zero-knowledge proofs, and multi-party computation. The core innovation is a co-processor model: rather than embedding privacy at the consensus layer, encrypted computation runs off-chain with on-chain verification. This keeps the base chain lean while enabling multiple parties to compute jointly over private data without a trusted coordinator. Auryn walks through the audit pipeline, testnet timeline, and three concrete primitives the team is targeting: secret ballot voting with public verifiability, sealed-bid auctions where bids remain hidden until close, and private AI training that learns from data it never sees in the clear. The discussion is grounded in engineering trade-offs — gas costs, proof generation latency, threshold assumptions for MPC — not whitepaper abstractions. Listeners walk away with a mental model for how encrypted execution fits into existing stack architectures and where the sharp edges still remain.
Key Insights
- Zodiac's transition from DAO-specific tooling to a general-purpose modular access control suite — an adaptation driven by real-world deployment friction rather than theoretical design.
- The Interfold's co-processor model deliberately avoids consensus-layer entanglement, keeping encrypted execution isolated and auditable without bloating the base chain.
- Combining fully homomorphic encryption (FHE) with zero-knowledge proofs and multi-party computation lets multiple parties compute jointly over private inputs while each retains sovereignty over their own data.
- Concrete testnet launch timeline and audit pipeline for encrypted execution infrastructure — moving from white paper to production with named milestones and security reviewers.
- Three real-world primitives: secret ballot voting (private tally with public verifiability), sealed-bid auctions (bids remain hidden until close), and private AI training (model learns without seeing raw data).
- The architectural bet against a single 'trusted coordinator' — distributing trust across cryptographic primitives rather than relying on any one operator or committee.
Who should listen: Infrastructure engineers and protocol architects evaluating how to integrate privacy-preserving computation into existing on-chain architectures without adding trusted third parties or consensus-layer overhead.
Why This Matters
The Interfold's co-processor approach aligns with a broader infrastructure pattern we track: offloading specialized computation (in this case, encrypted execution) to dedicated layers rather than forcing every use case through general-purpose consensus. This separation of concerns mirrors how rollups decoupled execution from consensus, and it signals where the next wave of privacy tooling will land — as composable modules, not monolithic chains.
