ANBD-GENESIS: The decision engine that knows when NOT to decide alone

Hey Product Hunt! 👋 I built ANBD-GENESIS because most "AI decision" tools force a false choice: automate everything (risky) or review everything by hand (doesn't scale). The engine explicitly refuses to decide when it doesn't have enough evidence — it escalates to a human instead. That's not a limitation, it's the core design. Under the hood it uses Vector Symbolic Architecture (binary hypervectors) instead of an LLM call, so it runs in ~4ms and costs a fraction of a typical AI API call — which matters when you're evaluating thousands of orders or transactions per minute, not chatting one at a time. Would love your feedback, especially from anyone dealing with order/refund fraud review at scale. Happy to answer any technical questions below!