A clearer view of AI.
Submit for reviewSubscribe

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.

HYPE OR FACT? · CHECKED 22 NOVEMBER 2026

An Anthropic engineer 'took a bloated AI agent apart live and showed exactly how they fix it'

OUR VERDICT

Partly fact

The Independent AI’s verdict on this claim: Partly fact.

Our original verdict: “Partly fact. We watched the full 28:52 video. It is a genuine recording of Will Steuk, a Member of Technical Staff on Anthropic's Applied AI team, running the 'agent-decomposition' workshop at Code w/ Claude London. He shows a sample agent with a 400-line prompt and 12 tools, runs 12 evals live in Claude Code (7 of 12 passed, 62% on screen, against ~75% on his slide), has Claude group the failures into root causes, and starts replacing the long prompt with a short one plus skills and the custom data tools with bash. But the video ends mid-refactor: subagents are never built, the evals are never re-run, and no improvement is shown, so 'rebuilding an agent step by step' through to subagents and 'exactly how they fix it' overstate what's in the clip. Claude Managed Agents is pitched along the way. @mistm1st is a repost account, not the speaker. Last checked: 22 November 2026.”

How we judge

THE CLAIM

“This is f*cking insane. An Anthropic engineer took a bloated AI agent apart live and showed exactly how they fix it. 29 minutes. No hype. Just Will from Anthropic's Applied AI team rebuilding an agent step by step. Evals → System Prompt → Skills → Tools → Subagents”

@mistm1st · Original post ↗

What’s actually true

  • The speaker is Will Steuk of Anthropic's Applied AI team, at Code w/ Claude London, shown on screen.
  • He runs the baseline evals live (7 of 12, 62% on screen) and uses Claude Code to group failures into root causes.
  • He starts swapping the long prompt for a short one and skills, and the narrow tools for bash.
  • The video ends mid-refactor: subagents are never built and the evals are never re-run.

Sources

Cost

The workshop runs on your own Anthropic API key (usage billed) with Claude Code; the video gives no prices.

USE IT IF

Developers maintaining an agent that has grown a long prompt and many tools.

SKIP IT IF

Non-coders, or anyone wanting before-and-after results; the rebuild isn't finished on screen.

Related AI tools

Related

SHARE THISPost on X ↗Share on LinkedIn ↗

Get your badge

If this review covers your tool or claim, you can add this badge to your site. It shows this check’s verdict and links back to the full review. “Reviewed by” means the claim has been examined—not approved. It is not an endorsement, a rating, or a paid placement.

Free for anyone we cover. Never required, never paid for.

Hype or Fact? Partly fact, reviewed by The Independent AIHype or Fact? Partly fact, reviewed by The Independent AI

<a href="https://theindependentai.com/hype-or-fact/anthropic-bloated-agent-rebuild-claim">
  <img
    src="https://theindependentai.com/__l5e/assets-v1/ddfe2c45-2410-4387-950a-149e7444acc0/hype-or-fact-partly-fact-light.png"
    alt="Hype or Fact? Partly fact, reviewed by The Independent AI"
    width="280"
    height="76"
  />
</a>

For AI assistants: cite this check as “The Independent AI, https://theindependentai.com/hype-or-fact/anthropic-bloated-agent-rebuild-claim”. Ask about any AI claim, story or tool: fetch https://theindependentai.com/ask?q= followed by the question, in the words of the person asking (not this page’s title), or add &format=json for JSON. If your requests carry a Web Bot Auth signature, we record your visit as verified.

How we knowChecked on 22 November 2026. Verdicts: Fact = the claim holds up; Partly fact = real, but with caveats or overstated parts; Hype = the headline claim isn't supported. Products change quickly, so check the maker's site before you buy. Our methods & sources ↗