In plain English: Quantum Brain is the cognitive substrate the other agents call — a 7-circuit reasoning engine that scores options, classifies inputs and makes decisions, amplifying the best answer with a real quantum step. Every reasoning trace is sealed, so a decision is never just "the model said so."
When an automated system makes a call that matters, "trust the black box" doesn't survive an audit, a dispute, or a board meeting.
When two options are nearly tied, a naive system picks one and moves on. The right move is to know it's close — and escalate.
The inputs, the weights, the chosen path — gone after the answer ships. Without a sealed trace, you can't show why you decided what you did.
Three run genuinely quantum in production via a calibrated router — GPU exact statevector by default, real IBM Heron on escalation. The rest are deterministic composites, labelled so.
A working brain_decide console. Give it options and a fit for each — it scores them, applies one Grover amplitude-amplification step to the leader, and returns the decision with its amplified confidence. If that confidence stays below 0.80, the call is escalated to a human, exactly as the live agent does. The reasoning trace is sealed.
brain_entangle, brain_walk, brain_dream — execute real quantum circuits via a calibrated router (GPU exact statevector, real IBM Heron on escalation). Five more carry real, dashboard-verifiable IBM Heron job IDs and run deterministic emulation in production today. Seven composites are pure audit-reproducible functions — not quantum, labelled so. All nine cited Heron job IDs resolve to real backends, status DONE, simulator:false. We cite no job ID that does not resolve.Add Quantum Brain to your watsonx Orchestrate workspace, or let the Prompt Builder wire it into a score-decide-seal chain.