my AI. / Demos
Key Improvements · Can Agents Be Trusted? · Council-validated
▶ A 10-minute technical explainer. Every claim was validated with a Karpathy-style multi-model Council against dated sources. Press play — slides advance automatically.
ℹ️ How this was made · voices · sources

What: An animated slide-explainer for technical teams on how to build agentic assets that earn trust — guardrails, LLM-as-Judge, the Karpathy Council, Automated Reasoning, and information-classified human-in-the-loop.
Hosts: Byron Arnao (Principal Technologist, Global Responsible AI, AWS) & Mia Mayer (Applied Scientist, AWS Machine Learning University). AI-generated voices (Gemini TTS), one consistent voice each.
Grounding: Drafted from Byron's podcast corpus — Nate B. Jones especially (information classification, judgment artifacts, agent control surfaces) — plus primary AWS sources.
Validation: Karpathy-style LLM Council (Gemini 2.5 Flash + Gemini 2.5 Pro → live web re-verification on knowledge-cutoff flags → Opus chairman). Unsourced claims cut; recent products verified against live AWS sources. Report: council-report.json.
Built by Byron Arnao's OpenClaw agent fleet.

my AI · Demos · Key Improvements · earntrust.demo.arnao.ai — Council-validated · Earn Trust.