AI for Software Development: A Reality Check
summary
Shows a working Thoughtworks sandbox (code on GitHub) that runs reusable prompts for epic breakdowns and STRIDE threat modeling, pulls in a 2,000‑page Wiki PDF for a medical records project, and uses retrieval augmented generation to surface relevant docs. Interaction patterns include one‑shot prompts, QA-driven story refinement that raises confidence and reduces back‑and‑forth, and a matrix view for risks. Notes over 50 clients using coding assistants (including Mercedes‑Benz) and cautions that curation and user experience still determine usefulness.
AI-written summary · about
chapters (11)
- 0:00AI in software delivery
- 1:05AI role and focus
- 4:57LLM assistance applied
- 7:46Building AI assistance
- 13:07AI interaction patterns
- 19:25Domain context for AI
- 22:12Retrieval augmented generation
- 27:03Knowledge firehose challenges
- 31:51Responsible context orchestration
- 33:17Mindset shift with AI
- 35:05Q&A
related talks
39mAI Assistance Beyond Code: What Do We Need to Make it Work?
27mAI for software development: A reality check
15mAI's Impact on Development, Post-Production & Support Journeys
48mAI for Teams: AI Assistance beyond Coding
27mAI-First Software Maintenance & AI-First Software Delivery
30mThoughtful AI for the Rubyist
26mAI-augmented SDLC: Rethinking how we deliver software
01h 02mSkills for the age of AI developer tools