03-10-2026Price:

The Frontier

Your signal. Your price.

AI & TECH

The Rise of AI Self-Improvement and Open Source Tools

Tuesday, March 10, 2026 · from 1 podcast, 3 episodes
This Week in StartupsThis Week in StartupsThis Week in Startups
  • Andrej Karpathy's Auto Research allows AI models to self-improve incrementally, showing self-improvement is viable.
  • Shopify's Tobi Lütke demonstrated a 19% performance boost, indicating wider access to AI experimentation.
  • Open-source projects like OpenClaw capture significant grassroots enthusiasm, contrasting with a negative perception of AI in the U.S.

AI is rewriting its own rules, and the tools are increasingly available. Andrej Karpathy’s Auto Research offers a stripped-down training loop enabling small AI models to iteratively enhance their own code in five-minute cycles. The practical implementation of this concept is not theoretical. It’s happening now.

Shopify CEO Tobi Lütke tested this model over a weekend, running 37 experiments that resulted in a 19% performance gain on an 800-million-parameter AI. This shift empowers not just researchers but any tech-savvy leader. Suddenly, the field of AI is opening up from a niche accessed by a few thousand individuals to thousands more who can now experiment and innovate.

A stark contrast in AI sentiment emerged globally. In China, the grassroots embrace of tools like OpenClaw is evident, fueled by local governments and enthusiastic meetups. Conversely, the U.S. grapples with skepticism; a recent NBC poll revealed only 26% of respondents support AI technology. The democratization of AI development is happening alongside a troubling public trust vacuum.

OpenClaw, another standout project, has outperformed mainstream incumbents by rapidly accumulating GitHub stars. In just 39 days, it became the most-followed open-source project, marking a significant shift in developer focus. As traditional companies chase new features, outsider projects like OpenClaw find ways to resonate more with the developer community.

The implications are vast. The advancements in self-improvement and the explosive grassroots adoption signify that the conversation around AI is evolving. Companies that once controlled the narrative are now contending with an insurgent wave of open-source innovation.

Andrej Karpathy, via This Week in Startups:

- It's a really stripped down LLM training loop and it runs in five-minute increments.

- So you bring your own AI model to be an agent essentially and then you give it a prompt and then what the system does is try to improve its own code over a five-minute training period.

Self-ImprovementOpen SourceGlobal Sentiment

Source Intelligence

What each podcast actually said

This Week in Startups
This Week in Startups

How agents will change banking forever | E2260 • Mar 10

  • Andrej Karpathy's Auto-Research tool enables an AI model to iteratively test and improve its own code in five-minute cycles, demonstrating a basic mechanic of self-improvement.

Also from this episode:

Models (3)
  • Shopify CEO Tobi Lütke used Auto-Research to run 37 experiments over eight hours, boosting a model's performance score by 19%, despite having no machine learning research background.
  • Jason Calacanis predicts AI tool democratization will expand the pool of people capable of improving models from roughly 3,000 highly-paid PhDs to hundreds of thousands of tinkerers.
  • Calacanis argues that elite AI labs are likely advancing similar self-improvement techniques at a pace twice as fast as the public tools indicate.
Society (2)
  • A recent NBC poll found only 26% of Americans view AI positively, with 46% opposed, indicating lagging public enthusiasm compared to technical progress.
  • The hosts contrast US skepticism with Chinese AI enthusiasm, where OpenClaw meetups draw crowds and local governments offer adoption incentives, driven by aspirational culture and tangible career utility.
Enterprise (1)
  • The barrier for non-technical executives to directly tinker with AI training loops has collapsed, foreshadowing tension with developers who prefer keeping management away from the codebase.
AgentsModelsBig TechStartups

Wisdom of the $TAO: the future is decentralized AI • Mar 6

  • The system monetizes open-source contribution in a way traditional development cannot, according to Jeffrey.

Also from this episode:

Mining (2)
  • Bit Tensor uses a crypto incentive layer with token emissions akin to Bitcoin mining rewards to subsidize AI development, according to guest Mark Jeffrey.
  • Jeffrey describes the model as Bitcoin's incentive structure applied to stranded talent instead of stranded energy.
Startups (7)
  • The network operates 128 specialized AI subnets that compete to produce the best models.
  • Ridges costs 29 dollars per month while centralized competitors raised funding at valuations in the billions.
  • The Ridges project was built on roughly 10 million dollars in chain emissions, compared to traditional startups requiring billion-dollar valuations.
  • Developers anywhere can earn subnet tokens daily by outperforming centralized teams, turning the stranded talent problem into a market.
  • A developer in Turkey can earn subnet tokens daily by improving the model, effectively owning a slice of the product's success.
  • The market bypasses traditional startup machinery including HR, payroll, and fundraising.
  • The network pays for progress directly, turning AI development into a performance-based contest.
Coding (2)
  • Subnet 62 launched Ridges, a coding assistant that scores 73 to 88 percent on benchmark tests measuring vibe coder effectiveness, according to Jeffrey.
  • Ridges scores competitively with Claude and Cursor on performance tests.
AI & TechMarkets

Is Anthropic Making the Biggest Mistake in AI History | E2258 • Mar 5

  • OpenClaw accumulated more GitHub stars than React in 39 days, becoming the most-followed open source project in history.
  • OpenClaw, an open-source coding agent, dethroned React as the most-followed project on GitHub in just over a month.
  • OpenClaw briefly partnered with Venice AI, an uncensored chat platform founded by crypto veteran Eric Vorhees.

Also from this episode:

Agents (1)
  • AI incumbents focused on 'agent' features and co-work tools, while OpenClaw captured developer mindshare by shipping code, according to the summary.
Startups (1)
  • Logan Allen of Finn Capital described OpenClaw's rise as an outsider project capturing developer attention while established players looked elsewhere.
AI & Tech (3)
  • Eric Vorhees applied blockchain-era principles, including user sovereignty, privacy, and censorship resistance, to the AI landscape via Venice AI.
  • Eric Vorhees, from the crypto world, observed that principles like user sovereignty, privacy, free speech, and lack of censorship were absent in AI.
  • Vorhees founded Venice AI to bring user sovereignty, privacy, free speech, and censorship resistance to the AI landscape.
Culture (1)
  • Jason Calacanis described a tech adoption curve starting with criminals, moving to discreet uses like sports wagering, then to mainstream users seeking efficiency.
Stablecoins (1)
  • Calacanis noted stablecoins are entering their final adoption phase, with users like gardeners asking for USDC payments to avoid 3% fees.
AI & TechCoding