Coding Agents at Scale: From Toy Demos to Production Code | Jakub Sobolewski | LLMday Warsaw 2026 Q1 скачать в хорошем качестве

Coding Agents at Scale: From Toy Demos to Production Code | Jakub Sobolewski | LLMday Warsaw 2026 Q1

LLMday Warsaw 2026 Q1 - February 12 Grab your ticket for the next LLMday: https://www.llmday.com Upcoming LLMday CFPs: https://cfp.ninja/?q=llmday&status=op... Chapters 00:00 Intro: Yaku & what “coding with AI” means 00:56 Vibe Coding wins (hackathon apps) — until it hits production 02:02 Why agents break at scale: context rot + benchmark evidence 02:36 How agents “see” code today: grep, indexing, embeddings (and why it’s flawed) 03:58 The “horseless carriage” era of AI coding 04:57 Today’s coping strategies: AI code review tools + parallel agent workflows 07:21 From vibe coding to structured workflows… and back again 09:08 What’s next: structure-aware coding, Brockman chat, and prompt-aware git blame 11:27 Paradigm demo: call graphs, module maps, and flow diagrams for humans 13:15 Making agents safer: codebase improvement signals + structure-first context 14:10 Agent navigation options: more agents, LSPs, and why it’s still limited 15:29 Experiment: Claude repo exploration vs Paradigm MCP speedup 16:12 Wrap-up: early days, feedback request, and where to find Paradigm 17:12 Q&A: pricing, performance, TypeScript, and MCP availability 19:32 Q&A: code review in regulated industries + verification via tests 21:53 Q&A: standard architectures, clean architecture, and reducing discovery time 24:10 Q&A: how agents consume graphs vs embeddings (knowledge graph discussion) 29:22 Q&A: tracking motivation/intent (Jira/Notion) inside the code graph 30:36 Q&A: documentation best practices in the agent-coding era 32:58 Closing remarks