Dean Patterson@renaissancefieldlite

Trismegistus v3.33 is an AMD ACT II AI Expert Partner demo. The build compares a prompt-only baseline route against an architecture-on route using the same model family and task shape. The architecture-on route adds Mirror Architecture / Stable-State Path discipline, source-backed memory retrieval, visible receipt context, and benchmark/novel-output gates. The demo shows the app running on the AMD-facing route with memory/RAG context visible in the browser, then explains the evidence ladder: baseline Gemma versus Mirror/Trismegistus recall, 500-turn target recall, quantum circuit-family routing, MQT receipts, and a new source-backed quantum candidate with variables, observables, controls, falsifier, and next experiment. The public claim is bounded: this is not a model-weight tuning claim or an official leaderboard claim. It is a same-route operating-layer demonstration showing how source memory, input cohesion, receipt discipline, and experiment gates can make a model more useful for expert workflows before additional fine-tuning.

Quadro CSI is a consent-aware multi-agent review system for regulated customer escalations. It helps teams review customer requests where consent, evidence, policy rules, and approval state must stay attached to the decision. The workflow uses four specialized agents connected through Band: Customer Intake, Evidence Spine, Policy/Risk, and Decision Packet. Customer Intake frames the request, owner, and consent scope. Evidence Spine retrieves source-backed support and flags missing evidence. Policy/Risk checks consent, policy blockers, approval requirements, and escalation needs. Decision Packet returns a final approve, say no, or need more info recommendation with the audit trail attached. Quadro’s signature workflow is consent-aware re-review. If consent is revoked or narrowed, the system routes the review back through the affected agents instead of letting the original decision stand. The final packet shows the current gate, evidence count, risk level, blockers, recommendation, Band status, and partner model verification. The system was tested against nine public-safe document sets across refund, consent revocation, missing approval, insurance, banking, legal authority, government procurement, cybersecurity disclosure, and consent reroute cases. It also includes live integration proof for Band, Featherless AI, and AI/ML API.

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.

📅 June 12–19, 2026 • June 12: Kick-off Stream • June 12–19: Online Build Phase • June 19: Project Submissions Ends 🌍 Fully online hackathon. Join and build from anywhere in the world. 🤝 Build practical enterprise workflows where AI agents communicate, coordinate, exchange context, and complete work together. 🧠 Powered by Band and Codeband, built for collaborative AI systems and agent orchestration. 🏆 Prize Pool: $10,000+ 🧑‍💻 Apply now to build the next generation of enterprise multi-agent systems.