AI for software development: A reality check
summary
Claims what works now: inline coding autocomplete and IDE chat with GitHub Copilot, natural‑language query builders like Google Cloud's Do it AI and Honeycomb, plus prototype integrations with Sourcegraph and Bloop. Evidence: coding is ~32% of dev time, some tasks show 55% faster task completion, and optimistic cycle‑time math yields ~133% time saved in coding portion only. Demonstrations include generated tests, SQL/prometheus query translation, doc summarization and story drafting that ask clarifying questions.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:47The hype cycle
- 1:22The two extremes
- 1:55Whats new
- 2:26Raising abstraction level
- 5:11Coding assistance
- 10:51Query translation
- 12:32Documentation
- 15:34On the fly documentation
- 17:22Data quality
- 17:57Writing user stories
- 20:35How can this impact delivery performance
- 22:01How can this impact cycle time
- 24:37Collaborative system
- 25:08Conclusion
related talks
24mAI and Web Development: Hype or Reality
01h 16mAI in Software Development: Boosting Efficiency
33mAI assisted software delivery
51mAI-Assisted Coding: The Future; Between Challenges And Benefits
41mBeyond Code: The Essential AI Skillsets for Development Teams
18mHumans + AI: The Future of Software Development Teams
15mAI's Impact on Development, Post-Production & Support Journeys
30mThe AI Revolution in Coding