← все видео

You are Not Ready: Agentic coding in 2026

Jo Van Eyck · 2026-02-08 · 31м 5с · 31 929 просмотров · YouTube ↗

Топики: ai-loop-engineering

Аудио ещё не скачано.

📝 Summary

Summary ещё не сгенерён.

📜 Transcript

Transcript ещё не сделан.

⚙️ Pipeline jobs

StageStatusAtt.UpdatedError
download skipped 0/3 2026-07-20 11:43:48 pruned: interest_score < 8.0

📄 Описание YouTube

Показать
AI coding tools are crossing a threshold. Not because of AGI, but because of long-running agents, memory outside context windows, and multi-agent orchestration. Individually these look like rough prototypes. Together, they point to a fundamental shift in how software gets built.

In this video, I break down what’s been happening in the AI-augmented coding space over the last few months: model improvements, long-running agent loops, task memory systems, multi-agent coordination, and why tools like Claude Code are early indicators. We are rapidly moving from “babysitting” AI to defining work, setting guardrails, and stepping out of the loop while agents execute.

I introduce a maturity ladder for AI-augmented software engineering, from basic chat usage to full multi-agent orchestration, and explain why you can’t skip any steps. Along the way, we look at which skills are fading (syntax memorization, boilerplate coding, manual refactoring) and which are becoming more critical: system design, problem framing, judgment, context engineering, and building agentic pipelines.

In the next coming months you’ll write less code, but you’ll take on more responsibility. The software engineering job isn’t disappearing, but it’s changing. And the shift is already underway.

📧 Stay up-to-date with my weekly agentic engineering newsletter: https://agentic-engineering-weekly.ghost.io/

Ralph: https://ghuntley.com/ralph/
Beads: https://steve-yegge.medium.com/introducing-beads-a-coding-agent-memory-system-637d7d92514a
Gas Town: https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16dd04
Openclaw: https://openclaw.ai/

Timestamps
00:00 Introduction
02:04 New models: specializing on long-horizon agentic workloads
03:06 Ralph Wiggum: while-loops around your agents
06:17 Beads: trello for your agents
07:36 Gas Town: multi-agent orchestration
08:31 Openclaw: your personal AI assistant
10:10 Claude code: incorporates all these ideas
14:25 AI coding maturity ladder and future skillset
30:38 Recap