HomeEverythingEducation
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
Story archiveAll categories
← All Stories

Anthropic uses human engineers to refine Claude Code

Created at 1 Jun · 5:16 AM3 sources↑ Market-relevant3 events
IN SHORT

Anthropic is employing human software engineers, paid up to $280 per hour, to improve its Claude Code AI tool. The project, codenamed 'Marlin' and managed by Snorkel AI, aims to train the AI to mimic professional developer skills by reviewing and refining code outputs.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Key Numbers

$280per hour contractor pay for coding project
1,000human software engineers providing feedback
13%Snorkel AI workforce reduction in September
$100 millionSnorkel AI Series D funding
$1.3 billionSnorkel AI valuation in May 2025

Who's Involved

Anthropic
AI company developing Claude Code
Snorkel AI
Data labeling startup facilitating AI training
Claude Code
AI coding tool being improved
Human software engineers
Contractors providing feedback and training
Anthropic uses human engineers to refine Claude Code

↳ Why This Matters

As AI models become more sophisticated, the need for specialized human input to refine their capabilities grows. Projects like Anthropic's Claude Code improvement highlight the increasing reliance on human expertise to train AI for complex tasks, moving beyond general data labeling to specialized fields like software engineering. This trend underscores the evolving landscape of AI development, where human oversight and domain-specific knowledge are crucial for creating reliable and high-performing AI systems.

Key facts

  • Anthropic is using human software engineers to improve its Claude Code AI tool.
  • The project, codenamed 'Marlin', aims to train Claude Code to mimic professional developer skills.
  • Contractors are paid up to $280 per hour for prompt creation and code review tasks.
  • The process involves A/B testing code from different models to ensure simplified, maintainable output.
  • Snorkel AI, a data labeling startup, is managing this work for Anthropic.

As AI models become more sophisticated, the need for specialized human input to refine their capabilities grows. Projects like Anthropic's Claude Code improvement highlight the increasing reliance on human expertise to train AI for complex tasks, moving beyond general data labeling to specialized fields like software engineering. This trend underscores the evolving landscape of AI development, where human oversight and domain-specific knowledge are crucial for creating reliable and high-performing AI systems.

Frequently asked questions

Project Marlin is an internal name for Anthropic's project, managed by Snorkel AI, that uses human software engineers to fine-tune the Claude Code AI model.

Contractors are being paid up to $280 per task, with each task taking approximately one hour, to create prompts and review code for Claude Code.

The goal is to improve Claude Code's performance by mimicking professional developer skills, ensuring it can produce detailed, simplified, and maintainable code.

The tasks require software engineering backgrounds, with Snorkel AI often working with individuals holding advanced degrees or equivalent experience.

What Happens Next

01The project is ongoing.
02Contractors continue to provide feedback and evaluate model outputs.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

1 Jun · 9:00 PM
Anthropic has filed for an IPO in New York, potentially valuing the AI firm at over $1 trillion.
Daily Mail via PiQSuite
1 Jun · 9:00 AM
New article discusses the broader trend of "efficiency-maxxing" in AI use, prompted by concerns over ROI and high token spending, impacting companies like Amazon and Uber.
Business Insider via PiQSuite
1 Jun · 4:02 AM
Anthropic is using feedback from about 1,000 human software engineers via Snorkel AI's Project Marlin to improve Claude Code's performance.
Business Insider via PiQSuite

Sources

T1
Inside the unseen operation to turbocharge Claude Codem.piqsuite.com
T1
Tokenmaxxing is entering its efficiency-maxxing eram.piqsuite.com
T1
AI firm Anthropic, maker of the popular chatbot Claude, fires starting gun on $1 trillion floatm.piqsuite.com

Related Stories

Elon Musk's Grok 4.5 Launched, Priced Aggressively Against Competitors
8 Jul · 10:30 PM
Nvidia CEO Jensen Huang: Engineers prefer building AI agents to coding
9 Jul · 5:35 AM
BofA extends $520 million credit line to OpenAI ahead of IPO
8 Jul · 4:59 PM
Google revamps Android AI dev benchmark, adds Fable 5 and other agents
8 Jul · 4:45 PM
AI Startups Report Accelerating Revenue Growth
8 Jul · 4:05 PM