AI Assistance for Software Teams: The State of Play
summary
Overview of current tooling: IDE-integrated coding assistants (notably GitHub Copilot as leader) give inline suggestions, integrated chat and RAG-style workspace indexing, and make adding code easy but refactoring weak — a Codesee study finds functionally correct refactorings only 37% of the time. Agent-style ticket-to-code features are experimental; a small migration agent (Enzyme → React Testing Library) shows promise but fails on complex cases. Test generation helps brainstorm edge cases but rarely automates legacy test backfills (author estimates ~99.9% require refactoring first).
AI-written summary · about
chapters (16)
- 0:00Intro
- 1:35Big picture: AI tooling archetypes
- 2:55Adjusted for size: Tooling maturity & current adoption
- 4:43Coding assistants: A tour
- 9:50Coding assistants: Core features
- 10:01Coding assistants: Model choices
- 12:43Coding assistants: Context orchestration
- 15:43Context of all your codebases
- 24:57Coding assistants: The next frontiers
- 25:20Testing
- 28:04Solving larger problems
- 30:12Code review & refactoring
- 33:31Documentation
- 36:41Debugging
- 42:29Final thoughts
- 44:48Outro
related talks
49mFrom autocomplete to agents: AI coding assistance state of play
33mAI assisted software delivery
50mAI Coding Assistance: The State of Play
35mGoing Beyond Copilot with AI Agents
26mFrom autocomplete to agents: AI coding assistance state of play
45mThe AI Revolution in Software Development: From Code Writing to Quality Assurance
27mAI for software development: A reality check
39mAI Assistance Beyond Code: What Do We Need to Make it Work?