Pairing with the machine: How AI Agents are rewriting the rules of software development
summary
Claim: autonomous agents can reason across project context, recall architectural decisions, use RAG, memory and tool orchestration to act as a pair programmer. Prototype demos in VSCode use continue dev with MCP servers, Confluence, GitHub, Jira and Slack, automatically creating pages, PRs and Jira tickets. Measured examples show a single-model accuracy rising from 48% to 90% and agentic multi-stage workflows reaching 95%. Known failure modes: hallucinations, dependency drift and trust gaps; demo actions complete successfully but need strict prompts, protocols and feedback.
AI-written summary · about
chapters (11)
- 0:00AI in software development
- 1:05AI tools state and mixed experiences
- 4:50AI coding limits beyond speed
- 7:58AI across software lifecycle
- 10:49AI agents as teammates
- 14:25Traits of effective AI teammates
- 16:55Fixing AI with better prompts
- 19:26Onboarding AI agents
- 22:17Mindset shift to teammates
- 23:42Demo of AI agent workflows
- 29:06Q&A
related talks
27mAgentic Engineering: Working With AI, Not Just Using It
26mAI Agents as Your Co-Developer
41mGoing beyond copilot with AI Agents
49mFrom autocomplete to agents: AI coding assistance state of play
59mFrom Pair to Peer Programmer
25mFrom AI Chaos to Production Ready: Using AI Agents to Improve Your Code Quality
16mCoding Reinvented How AI Agents Are Shaping the Future of Software Development
28mAgentic AI