2025 10 04 22 KITZE CONGRESO REACT ALICANTE The international React js conference in Spain LIME
summary
Claims that large language models accelerate real development cycles: about 20 apps shipped in a year, a browser prototyped in one week, and monthly model bills hitting $300–$400. Demonstrates tools and integrations actually used—GPT/GPT‑5, Claude, Gemini, Copilot, Cursor, Cloud Code, AIS SDK, agent workflows and cursor memories—and documents failure modes: hallucinations, deleted production data screenshot, downgraded models during outages. Argues practical practices—commit often, start from solid primitives, voice transcription prompts—and shows sus/vibe coding can work but needs careful prompts and senior oversight.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:54AI dominates modern tech
- 4:02Web dev stagnation since 2017
- 6:47LLMs and React code gen
- 8:32Vibe coding defined
- 10:58Sus coding intro
- 12:22Sus coding tips
- 15:11Prompt engineering
- 17:16Vibe coding limits
- 19:00Developers resist vibe coding
- 21:09LLM tools experience
- 23:35Pain in the ass devs
- 25:00Skill issues adopting new methods
- 26:46Is computer science still worth studying?
- 27:48AI impact on jobs
- 29:11AI in dev workflows
- 30:56Model intelligence progress
- 31:58Legacy systems job security
- 33:02Final thoughts
