Fine-tune the COE process with generative AI
summary
Lightning proof of concept shows generative AI turning human incident notes into a full COE document: a Jupyter notebook sends timeline and facts (10,000 files failed to transform) via Boto3 to Amazon Bedrock models (Anthropic Claude 3 and Sonnet), which generate a concise 200‑word impact, a zero‑shot five‑whys root cause path, and three SMART action items. Output enforces no blaming, produces an executive summary, and demonstrates reproducible prompts and API flow for faster remediation.
AI-written summary · about
chapters (12)
- 0:00Introduction to COE
- 1:07Why correction of errors matters
- 3:19Human challenges in COE
- 5:09Generative AI in COE
- 8:06Demo setup and inputs
- 10:40Prompt engineering basics
- 12:28Generating impact statement
- 14:17Five whys with AI
- 16:45Action items from AI
- 18:32Executive summary generation
- 19:38Human prompt engineering exercise
- 20:00Resources and closing
related talks
35mContinental Resources, PwC leverage generative AI to optimize oil & gas ops
24mAIGenOps: Integrating Generative AI into Platform Engineering
27mBuilding Generative AI features
31mGenAI after the Hype: Transforming Organizations with GenAI-based Agents
31mTesting, Evaluating and Debugging Generative AI and Agentic Ap...
29mGenerative AI: Path to Value in practice
21mRevolutionizing software creation: Harnessing the power of GenAI
53mThe Evolution and Impact of Generative AI: Transforming Creativity and Operations