A clearer view of AI.
Subscribe

WEEKLY NEWSLETTER

The AI Insider

A weekly round-up of the AI tools worth your time, and the hype to ignore.

Unsubscribe at any time using the link in each newsletter.

THE INDEPENDENT AI.

COMMUNITY PICK · 3 NOVEMBER 2026

Watch out: more recycled 'free courses' on harness, loop and graph engineering

3 November 2026, 00:24 · 3 sources · Curated by Sara Simeone

At least six accounts posted 'free courses' on harness, loop and graph engineering credited to Google, Andrew Ng or an 'OpenAI engineer'. They follow the same pattern of re-uploaded, re-captioned talks we've flagged since late October, and one is a re-post of an August video. A more useful line came from builder @Vtrivedy10, who argued that harness engineering is mostly a way of getting the right context to the model, so the context window is what you should design around; that's his opinion.

Why it mattersDon't share these 'courses' with your community. If you want to learn the topic, start by writing down what goes into the model's context at each step of one of your own workflows.

Curated by Sara Simeone as part of our human editorial coverage. The original sources are linked below; curation is distinct from a Hype or Fact? assessment.

Sources

How we knowCompany figures are labelled as their own claims; anything we couldn't verify is labelled as a report or a claim.

MORE FROM 3 NOVEMBER 2026

  1. FROM THE COMMUNITYAnthropic's IPO filing: it may 'imminently' hit its own threshold for AI-driven progress
  2. FROM THE COMMUNITY'Weight-space programming': GPT-6 Astra edited a small model's weights and lifted its maths score from 7.9% to 48.1%
  3. FROM THE COMMUNITYGDPx: the best AI agent finishes only 12% of 206 hard, high-value work tasks
  4. FROM THE COMMUNITYOpenAI's always-on agent, dots, is moving to GPT-6.1 Terra to reach more people
  5. FROM THE COMMUNITYAltman: the model behind OpenAI's Navier–Stokes result is coming 'very soon' (no date)
  6. FROM THE COMMUNITYHydrogen-1.0: an open-weight science model that its maker says beats GPT-6 Astra on research questions
All community picks ↗