Deterministic Rules Engine for Compliance | Dweve Lattice

What is Dweve Lattice?

Lattice is a deterministic policy and constraint evaluator for governed decisions. It compiles versioned rules into a checksummed, memory-mappable rulebook that runs inside the calling application, then returns a repeatable result with the rule version and owner boundary available for review. Source access is currently by request through Dweve's community.

• Replay uses the same rulebook version and recorded inputs. Lattice can show which rule ran, but it does not decide whether a policy is fair or certify legal compliance.
• The replay output is a decision record: it ties the rulebook version and recorded inputs to the result and the accountable owner boundary.
• Scoped benchmark results report 1.36 ns for the hot path, 39 ns for a cold lookup after mmap, and up to 735 million lookups per second on one thread under the referenced benchmark. Treat these as benchmark conditions, not a universal latency promise.
• The evaluator runs on hardware you control, without a network service in the decision path. Rule owners remain accountable for policy content and operational decisions.

For consumers

Lattice is a rules engine for decisions a person can check. It applies explicit rules to live data and gives the same answer under the same rules, so a decision can be explained later without technical guesswork. It publishes in the third round of Dweve's foundation release programme.

For businesses

Lattice is a deterministic policy and constraint engine in Rust for AI governance, compliance, pricing and access decisions. It compiles versioned rules into a checksummed artefact and evaluates live data on hardware you control, so every decision can be replayed and audited.

For engineers

Lattice is a deterministic constraint engine in Rust for compliance, pricing and access rules. Under the referenced benchmark, hot-path evaluation averages 1.36 ns and returns bit-exact results across supported platforms.