From autocomplete to agents: AI coding assistance state of play
summary
Overview focuses on concrete claims and demos: measurable speedups (optimistic 13% ballpark from earlier marketing but typical 5–12% observed), live agent behaviours that run tests, execute shell commands and fetch web fixes, and tools like Windsurf cursor client, Cloud Code, GitHub Copilot compared by capability. Examples include agents auto-fixing syntax, increasing memory (a brute force fix), introducing extra constructor params (spaghetti), and GitClear metrics showing more added LOC, higher AI churn and less refactoring — plus MCP server and supply‑chain risks.
AI-written summary · about
chapters (20)
- 0:00AI coding assistance intro
- 1:03AI coding tools evolution
- 3:12AI impact on productivity
- 5:18Supervised agentic coding intro
- 6:41How agentic coding works
- 8:26Agentic assistance examples
- 10:09Agentic assistance features
- 11:32Custom agent instructions
- 12:57Planning with AI best practices
- 14:23Expectations for agentic AI
- 15:26Common pitfalls and failure modes
- 17:10AI maintainability risks
- 18:35Code quality trends with AI
- 19:16Security risks in agentic environments
- 20:18Future AI code design
- 21:01AI tool responsibilities
- 22:06Debate on vibe coding
- 22:47Feedback loops and safety
- 23:09Team strategies with AI
- 24:13Balancing AI enthusiasm
related talks
49mFrom autocomplete to agents: AI coding assistance state of play
26mAI Agents as Your Co-Developer
50mAI Coding Assistance: The State of Play
30mHow AI Agents Will Change the Future of Programming
46mRecalibrate - How AI shakes up software team
30mBeyond the Hype: Using AI Coding Agents Today
16mCoding Reinvented How AI Agents Are Shaping the Future of Software Development
26mAI, You complete me