Inner I Residuals observes, evaluates, and secures every AI output across your stack — from pattern audits to policy enforcement. One coherent foundation. Three products.
Audits designs against 7 coherence constraints
Enforces policy on every agent action
Scores and signs AI outputs
Invariant observer · 7 coherence constraints · HMAC-signed audit trail
Every request passes through the same coherence infrastructure — no matter which product you're using.
Every AI request enters through the same gate. The Observer Core captures input, context, and intent — before anything is executed.
ψ₀ = invariant(observation)
The invariant observer vector anchors each request to a stable coherence baseline. Drift gets flagged. Consistency gets scored.
ψ₀ → coherence_score
Every output is tested against 7 coherence constraints. Pattern audit, policy check, output scoring — all the same engine.
score(ψ₀, constraints)
A signed receipt — HMAC-SHA256 — archives the result in the append-only chain. One audit trail, one coherence score, one source of truth.
receipt = sign(chain_id, score)
Each product is a different surface on top of the same Residuals Layer — so you don't have to build coherence twice.
Upload any design or UI screenshot. PatternFolio scores it against 7 coherence constraints — adversarial patterns, visual alignment, brand integrity. Built for teams shipping AI-generated designs.
Define policies. Evaluate every tool call. Maintain a tamper-evident HMAC-SHA256 hash chain. Inner I Secure brings enterprise-grade governance to AI agent deployments.
Every AI output passes through the Residuals scorer and exits with a signed coherence receipt. Proof of Awareness makes AI outputs auditable, verifiable, and accountable.
The SDK handles coherence checking, chain recording, and policy evaluation — so you can focus on building, not governing.
Full TypeScript SDK with type definitions. One import to rule them all — ψ₀ scoring, policy evaluation, chain operations, all in one client.
npm install @innerinet/residuals
View on GitHub →
Full REST API for policy management, chain inspection, and coherence evaluation. Works with any stack — Node, Python, Go, whatever you're running.
https://shapeos-3.polsia.app/api/secure
API reference →
The coherence architecture is grounded in published research. Start here if you want the full story.
Walkthrough of the coherence layer in action: how ψ₀ anchors, 7 constraints score, and signed receipts get archived.
Technical deep-dive: how the coherence filter works, why ψ₀ is invariant, and how truth compression improves AI reliability.
Accuracy, latency, and false-positive benchmarks across all three products. Full methodology and dataset documentation.
Benchmark running — see results →Audit your first design in 2 minutes. No sign-up required.
Launch PatternFolio →