Beyond Black-Box AI: Designing Transparent Automation for Regulated Industries
summary
Claims a concrete method called Chain-of-Work that pairs LLM reasoning with code-based execution to produce a full audit trail and reproducible runs, not just explanations. Platform is model-agnostic (Gemini Claude GPT), supports on‑prem and cloud, and targets regulated banks and utilities. A deployed trade‑finance flow drops from >2 hours to 9 minutes, supports 100+ business "citizen developers" with plans for thousands, and offers self-healing handling of messy data — hallucinations persist but the system routes around them.
AI-written summary · about
chapters (14)
- 0:00Transparent AI automation intro
- 1:07AI adoption in regulated industries
- 3:14Black-box AI limitations
- 5:41Traceability in regulated automation
- 9:13AI business risks and ROI
- 11:55AI meets RPA strengths
- 13:21Natural language automation
- 17:38Digital workers scalability
- 18:20AI computer architecture
- 19:24Trade finance automation
- 24:46Chain of work vs thought
- 27:58Automation results and benefits
- 29:05Enterprise compliance readiness
- 32:59Summary
related talks
24mFrom Black-Box to Guard-Box in Financial AI Security
20mBuilding Deterministic Infrastructure for Non Deterministic AI Agents
50mFrom Prompt Hacking to Architectural Determinism: Engineering Reliable GenAI Systems
13mDeploying Enterprise-Grade Compliant Generative AI
22mAgentic AI for Autonomous Compliance in Enterprise Tax System
32mBeyond LLMs: Building an AI-Powered Decision System at Scale
34mAI Agents: Transforming Autonomous Decision-Making
43mArchitecting the Future: Modern Solution Design with Agentic AI