AI InfrastructureCASE RECORD
Asadi Gate
Interactive governance demo — deterministic policy cascade for AI payment requests with SHA-256 evidence packs.
Operator impact
Proved that AI commerce governance can be deterministic and fail-safe — zero silent authorizations, every decision cryptographically receipted.
01PRIMARY SURFACEAI INFRASTRUCTURE
06 / ASADI GATEON RECORD

Asadi Gate — policy cascade evaluating a payment request
02BUILD SCOPE
Policy engine (8 rules)3 scenariosanimated rule cascadeserver-side receipt generationhuman approval queueaccessibility (reduced motion, screen reader, noscript fallback)
03PROBLEM / ROLE
Problem
AI agents initiating financial transactions need governance that is deterministic, auditable, and fail-safe — not probabilistic filters that silently approve when they break.
Role
- Architecture through deployment — policy engine, receipt system, full UI
- Designed the rule priority model and evidence pack format
- Built accessibility layer: screen reader, reduced motion, noscript
04APPROACH / OUTCOMES
Approach
- Designed and shipped a complete interactive demo: scenario selection → policy evaluation → verdict → evidence receipt
- Pure-function policy engine: 8 rules evaluated in strict priority order with zero side effects
- Fail-closed doctrine: if no rule matches or the engine errors, the transaction is blocked
- Human escalation for transactions exceeding agent authority — approvals expire in 72 hours, no auto-approve
- SHA-256 evidence packs generated server-side seal the full decision chain
- Animated rule cascade for transparency with prefers-reduced-motion support and full noscript fallback
Outcomes
- Zero silent authorizations — every decision path produces a verdict and a receipt
- Governance patterns from Truvesta, Ardura, and Equira applied to agentic payment flows
- Ships as a working product demo, not a slide deck — try it at asadi.ai/gate-demo
05PROOF LEDGER04 ENTRIES
- Pure-function policy engine: 8 rules, strict priority, zero side effects
- Full accessibility: reduced motion support, screen reader announcements, noscript fallback
- Fail-closed: blocked by default, receipted by design
- Client computes verdict, server generates evidence — trust boundary enforced
06REFUSED PATHSFAIL-CLOSED BY DESIGN
What I refused to build
These constraints are part of the engineering signal: the work stayed useful because the unsafe shortcut paths stayed out.
- R-01Auto-approve for escalated transactions — approvals always expire, never auto-pass
- R-02Client-side hash generation — server is the sole authority for evidence packs
07STACK / ACCESS
Stack
Next.jsTypeScriptSHA-256Policy EngineFramer Motion
Hiring signal
This is the kind of work I want ASADI AI to make legible first: fintech systems where product direction, architecture, execution, and governance all have to line up.