Ask Me Anything - AI – Tim Howard
summary
Defra reports moving policy compatibility checks from months to days by prototyping GenAI agents, proves delivery speedups (audited metric: 26 minutes per user from earlier GitHub Copilot rollouts and internal claims of up to five times faster code generation), and builds an AI SDLC playbook (https://defra.github.io/defra-ai-sdlc/) plus a code‑review tool that scans repositories, governance automation and an agentic framework for policy; a tiny enabling team of three runs hub‑and‑spoke adoption and measures outcomes.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:06Cross-organization IT challenges
- 2:32Defra’s GenAI journey
- 4:17Internal tools and governance
- 5:39GenAI playbook creation
- 6:43Measuring GenAI impact
- 7:46Human concerns about GenAI
- 9:21Q&A: Code maintainability
- 11:49Q&A: Refining AI requirements
- 13:15Q&A: GenAI for visualizations
- 14:38Q&A: GenAI sustainability
- 16:24Q&A: Engaging laggards
- 18:54Q&A: Sharing GenAI experiences
- 21:05Q&A: Mitigating hallucinations
- 22:58Q&A: Legacy code with GenAI
- 25:10Q&A: Training juniors
- 26:55Q&A: Team structure changes
- 28:19Closing remarks
related talks
01h 07mNew Methods of Working with AI.
46mRecalibrate - How AI shakes up software team
26mAI, You complete me
27mSoftware engineering with GenAI
39mAI as a Catalyst, Not a Replacement
57mOne Team’s Journey to Using AI Without the Hype
26mFrom autocomplete to agents: AI coding assistance state of play
24mAI Governance for developers: A practical framework for deploying autonomous coding agents safely…