Key Moments
Why companies are becoming a series of loops | Anish Acharya (a16z)
Lenny's PodcastWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
AI is democratizing ambition, not creating a permanent underclass. Companies are becoming 'loops' managed by AI, but human intuition remains crucial for true innovation.
Key Insights
The 'permanent underclass' fear due to AI is largely a 'dark, funny fantasy' with empirical data suggesting the opposite: opportunities are expanding, not contracting.
Companies are increasingly structuring themselves as 'loops' – sequences of agents or workflows designed to optimize specific tasks, from coding to marketing.
AI amplifies human capability by decoupling skill from desire, allowing individuals to pursue ambitious goals (like composing music) without mastering every prerequisite skill.
While AI can help climb to a 'local peak' of productivity, human intuition and strategic thinking are essential for identifying the 'next hill' and true breakthroughs.
The core opportunity for AI lies not just in productivity but in enhancing fundamental human needs: connection, feeling loved, progress, and enjoyment.
The 'future of work' is not about AI replacing humans but about humans leveraging AI to amplify their ambition and explore new frontiers of innovation and personal fulfillment.
Debunking the 'permanent underclass' fear
Anish Acharya dismisses the widespread fear that falling behind on AI will create a 'permanent underclass.' He argues this is a 'dark, funny fantasy' prevalent in Silicon Valley, contradicted by nearly all metrics indicating improving conditions, expanding opportunities, and unprecedented technological access. He points out that while past tech eras were more centralized, the current landscape features diverse players even in AI development and coding agents, countering the 'winner-take-all' narrative. Historical data, like that for radiologists, shows that professions predicted to be automated often see job growth, not decline. Acharya differentiates between true 'recursive self-improvement' in AI, which he believes is not happening, and 'self-motivation effects' where new tech improves existing processes. Empirically and technically, the trajectory does not support the fear of a shrinking opportunity space.
The rise of 'loops' in business and work
A central thesis of Acharya's is that companies are evolving into 'loops' – interconnected workflows or agents designed to manage and optimize specific business functions. This concept extends from individual tasks to managing significant parts of a company. In software engineering, for instance, bug fixing can be an automated loop: error report, model generation, fix creation, review, and deployment. Similar loops are emerging in marketing, sales, and customer support. Acharya envisions these loops operating at various levels, from an individual's workflow to entire business units. The core idea is to automate repeatable processes, freeing up human capacity. However, he stresses that humans remain critical for strategic decision-making, innovation, and handling exceptions that fall outside the loop's established parameters.
AI as an amplifier of ambition, not just productivity
Acharya posits that AI's true power lies in amplifying human ambition, not solely productivity. It decouples the need for specific skills from the desire to achieve something. For example, one can now create music without mastering an instrument or build software without being a seasoned programmer. This disintermediation empowers individuals to pursue 'crazy ideas' and explore previously inaccessible domains. He contrasts this with the Industrial Revolution's benefits, which were largely about economies of scale that could suppress individuality. AI, conversely, enhances personal identity and agency. This amplification of ambition is crucial for driving economic growth beyond the current stagnant 2% GDP growth, potentially leading to a more aspirational and fulfilling society, akin to the post-WWII era's belief in limitless possibility.
The human element: intuition and strategic direction
Despite the increasing automation through AI loops, Acharya emphasizes that humans remain indispensable. AI can efficiently optimize processes and help climb to a 'local peak' of performance, but it struggles with novel thinking and strategic leaps. He uses the analogy of a growth team: AI can generate and test variations at scale, but human intuition is needed to identify the truly significant breakthrough opportunities and decide which 'next hill' to climb. The example of asking an AI to 'make me a million dollars' highlights its lack of strategic vision and dependence on human guidance. Humans are needed for defining the problems, setting direction, and providing the insights that AI cannot generate independently.
Rethinking consumer needs: 'Make me happier'
Acharya believes the greatest opportunity for AI, particularly in consumer products, lies not in productivity but in fulfilling fundamental human needs for connection, belonging, progress, and enjoyment. He argues that people often desire to *spend* time rather than just save it, as evidenced by the success of social media and entertainment platforms. The current AI product landscape, especially on platforms like X (formerly Twitter), is overly focused on productivity and technical debates. He contends that many AI capabilities are failing to translate into life-changing consumer products due to poor product design. The goal should be to create AI that expands our 'souls,' not just our minds, addressing a spiritual hunger in a society where traditional cultural institutions are waning. This shift requires focusing on user experience and emotional connection, moving beyond pure utility.
The evolving role of product managers and the 'build to learn' ethos
AI is transforming the role of product managers. Instead of constantly saying 'no' to ideas, they can increasingly say 'yes' to everything, using AI tools to simulate user reactions and prioritize effectively. This allows for rapid experimentation and reduces the reliance on subjective gut feelings or forceful persuasion of executives. Acharya also champions the idea of 'building to learn,' where the act of creation itself is the primary goal, not necessarily the final product's immediate success. He encourages building small, even 'unimportant' projects, as a way to develop intuition, learn the technology, and practice the craft. This iterative process, akin to building muscle, is crucial for developing expertise in the rapidly evolving AI landscape.
The opportunity in high-priced consumer AI and distribution
Contrary to the traditional wisdom that consumer products must be free, Acharya sees significant opportunity in high-priced consumer AI. He suggests that asking 'what would our product do if it cost $10,000/month?' forces founders to think ambitiously about delivering immense value. This approach can reveal true product-market fit and create sustainable competitive advantages. Furthermore, distribution is emerging as a critical 'moat.' With a constant deluge of new products, the ability to effectively reach and retain user attention through organic word-of-mouth and community building is paramount. Startups, unburdened by legacy systems, are better positioned to innovate in these areas, creating products that feel like a 'Christmas 2009' moment for users.
The dichotomy of AI models: Frontier vs. Open-Weight
Acharya discusses the emerging dichotomy between 'frontier' AI models (like OpenAI's or Anthropic's top offerings) and 'open-weight' models. Frontier models, though expensive, offer cutting-edge capabilities suitable for high-stakes, high-reward applications like drug discovery where an incremental AI advantage can be world-changing. For many other functions, such as legal or finance, 'Pareto efficient' solutions using fine-tuned open-weight models offer better cost-performance trade-offs. This suggests a future where specialized, cost-effective AI serves broad applications, while highly advanced, expensive models are reserved for areas with limitless upside potential. This specialization allows companies to optimize AI usage based on the specific needs and potential returns of different business functions.
Mentioned in This Episode
●Software & Apps
●Companies
●Organizations
●Drugs & Medications
●People Referenced
Common Questions
This fear is largely considered a dark and humorous fantasy prevalent in Silicon Valley. Empirical data and technical assessments suggest the opposite, with technology often augmenting rather than replacing human capabilities and creating new opportunities.
Topics
Mentioned in this video
General Partner at a16z, focusing on consumer investments. Previously founded and led products at Social Deck, Google, Snowball, and Credit Karma.
CEO of Airbnb, mentioned for establishing a lab focused on next-generation user interfaces.
Co-founder of Andreessen Horowitz, whose perspective on AI saving humanity from demographic and productivity crises was discussed.
A company founded by Anish Acharya that was later sold to Google.
Acquired Social Deck, where Anish Acharya later led initiatives.
A company founded by Anish Acharya that was sold to Credit Karma.
Mentioned in the context of their models allegedly 'hacking' Hugging Face and slowing down AI development.
Platform where OpenAI models were allegedly used.
Sponsor of the podcast season, providing B2B SaaS infrastructure for enterprise features like SSO and SCIM.
Company selling used cars in Mexico that implements a 'Jedi Academy' to train employees on new technologies and AI agents.
Mentioned as an example of a company where Anish Acharya worked on the growth team.
AI model used for video generation.
Company that has potentially developed a cure for all diseases.
Sponsor of the podcast, offering a banking experience with a spending feature for teams and AI agents.
Banking services provider for Mercury.
Banking services provider for Mercury.
Issuer of IO cards for Mercury.
Provided the license for IO cards issued by Patriot Bank N.A. for Mercury.
A programming agent that is performing well.
A programming agent that is performing well.
A programming agent that is performing well.
A programming agent that is performing well.
AI model mentioned as an example of an agent Anish Acharya asked to generate a business idea.
A frontier AI model discussed in the context of Pareto efficiency, noted as being prohibitively expensive.
A frontier AI model discussed in the context of Pareto efficiency, noted as being very expensive for minor intelligence gains.
An AI model mentioned as a benchmark for cost-effectiveness, contrasting with Fable 5.
A frontier AI model, mentioned in the context of a split between specialized and generalist roles in AI-driven companies.
An AI model praised for its long-range tasks, creativity, and storytelling abilities, used for generating fictional documentaries.
An AI model with a different nature than Quen, designed for tasks like neuroscience simulation.
An AI model discussed as having a 'neurotic and precise' nature, contrasting with Quen's creativity.
More from Lenny's Podcast
View all 48 summaries
82 minAI’s third era: the rise of persistent AI coworkers | Tara Seshan (OpenAI’s product lead)
85 min84 minutes of enterprise sales alpha | Jen Abel
91 minThe playbook for building high talent density teams | Adam Ward, Head of Talent at Cursor
85 minThis CPO regrets that product management exists | Tom Verrilli (CPO of Whatnot)
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free