How AI Agents Will Change the Future of Programming
summary
Claims that agentic code assistants use tools to open and edit files, run unit tests, and search the web and follow a three-step loop of planning then reflecting rather than one-shot generation. Benchmarks (SSWBench verified) move from ~7% solved a year ago to ~30% months ago to over 50% now, with a projection past 75%. Zencoder ships a VS Code/JetBrains plugin; an agent fixed a UI flicker in minutes that took a senior developer two days. Bench tasks split: ~40% <50 minutes, ~50% ≈1 hour, ~10% >4 hours.
AI-written summary · about
chapters (9)
- 0:00AI agents in programming
- 1:07AI in coding tools
- 6:29What AI agents are
- 9:00AI agents transform workflows
- 12:14Evaluating with SWE-bench
- 16:12AI agents impact debugging
- 18:48Future of programming with AI
- 20:34Domain knowledge key
- 22:01Q&A
related talks
26mAI Agents as Your Co-Developer
16mCoding Reinvented How AI Agents Are Shaping the Future of Software Development
18mCoding Reinvented: How AI Agents Are Shaping the Future of Software Development
26mFrom autocomplete to agents: AI coding assistance state of play
35mGoing Beyond Copilot with AI Agents
49mFrom autocomplete to agents: AI coding assistance state of play
41mGoing beyond copilot with AI Agents
27mAgentic Engineering: Working With AI, Not Just Using It