Why companies are becoming a series of loops | Anish Acharya (a16z)

Summary of Why companies are becoming a series of loops | Anish Acharya (a16z)

by Lenny Rachitsky

1h 19m•September 6, 2026
Listen to Episode

Overview of Why companies are becoming a series of loops | Anish Acharya (a16z)

In this episode, Lenny Rachitsky interviews a16z consumer investor Anish Acharya about how AI is reshaping work, products, company design, and consumer behavior. Anish argues that fears of a “permanent underclass” are overstated, AI progress is more of a slow diffusion than a sudden takeoff, and the biggest opportunity is not just productivity—it’s building “loops” that help people and companies move from input to outcome faster, with humans still playing a critical role at the points where judgment, strategy, and out-of-distribution thinking are needed.

Key Takeaways

  • The “AI permanent underclass” fear is mostly a Silicon Valley fantasy.

    • Anish thinks the evidence points the other way: opportunity is broadly distributed, companies are adopting AI across functions, and job market data does not yet support mass displacement.
    • He argues the real change is economic diffusion, which tends to be slow.

  • AI is better seen as an amplifier of agency than a replacement for humans.

    • It “unbundles skill from desire,” enabling people to create music, software, products, and businesses without mastering every underlying craft.
    • The technology expands ambition as much as productivity.

  • Companies are increasingly becoming “series of loops.”

    • AI systems can take inputs, generate outputs, check results, and iterate.
    • This pattern is spreading from engineering into sales, support, marketing, legal, and operations.

  • Humans remain essential where the loop plateaus.

    • AI can help a team climb to a local maximum, but then human intuition is needed to identify the next hill.
    • The best org design is a back-and-forth between automated loops and human judgment.

  • Consumer AI’s biggest opportunity may be emotional, not just functional.

    • Anish believes people often want to spend time, not save time.
    • He sees major opportunities in products that help people feel:
      • more connected
      • more loved
      • more fulfilled
      • more fun
      • more progress in life

AI, Jobs, and the Future of Work

Why he’s not worried about mass unemployment

  • Most companies will adopt AI rather than replace humans outright.
  • Executives want to grow more ambitious businesses, not simply make the same business slightly more efficient.
  • Many jobs are not purely intelligence-bound; they involve physical constraints, customer context, and judgment.

What changes inside companies

  • AI is more likely to:
    • compress roadmaps
    • increase output per person
    • reduce admin and busywork
    • let people focus on their highest-value work
  • Anish cites examples where teams are using AI to rip through roadmaps much faster rather than lay people off.

What humans still do best

  • Sales
  • Strategy
  • Exceptions
  • Taste and judgment
  • Knowing what the model doesn’t know

The “Loop” Framework for Company Building

The core idea

Anish’s thesis is that companies will increasingly use AI to automate functional loops:

  • Engineering loop

    • bug report → repro → fix → review → ship → customer notification

  • Growth loop

    • generate variants → test them → analyze results → ship winner → preserve holdout → repeat

  • Business loop

    • signals from each function feed into higher-level decisions about product, pricing, strategy, or even business model changes

Why loops matter

  • AI is especially good at verifiable tasks and repeatable workflows.
  • The loop helps teams climb to local maxima quickly.
  • Human insight is needed to redefine the problem when incremental improvements stop paying off.

A practical mental model

  • Whenever a model gets stuck, ask:
    • Is this a knowledge gap?
    • Is this a data gap?
    • Is this a judgment gap?

Then feed the agent better context or involve a human coach.

Consumer AI: The Biggest Opportunity Areas

Anish sees three major consumer categories:

1. Coding agents as general-purpose builders

  • Coding agents are becoming a way to interact with the world, not just write code.
  • People are already using them for:
    • video editing
    • game creation
    • mini-app building
    • personal automation

2. Personal agents

  • He’s excited by products like GrokBot, ChatGPT’s agentic modes, and other assistant-like tools.
  • The winners will likely feel like a true assistant that knows your context, your threads, and can actually get work done.

3. Entertainment, companionship, and creative tools

  • This includes AI products that are emotionally engaging, socially interesting, or creatively expansive.
  • He thinks this area is under-discussed because it’s uncomfortable, but potentially huge.

Model Selection and “Model Sommelier” Thinking

Not all models are the same

Anish strongly disagrees with the idea that models are commodities. He says different models have different “shapes”:

  • some are more creative and open-ended
  • some are more precise and neurotic
  • some are better at long-horizon tasks
  • some are more cost-efficient for bounded jobs

His advice

  • Use models constantly.
  • Build small projects with every new model release.
  • Learn by shipping, not by reading release notes.

His practical approach

  • He keeps a chassis of projects and experiments.
  • He tries to use each new model to make something real.
  • This builds intuition about where each model excels.

Moats, Distribution, and Why Ambition Matters More Now

Moats are often discovered, not designed

  • Anish says startups often don’t know their moat upfront.
  • Durable advantages emerge from:
    • usage
    • network effects
    • brand
    • data
    • compounding product craft

Distribution still matters a lot

  • But he thinks many “distribution problems” are really product problems.
  • If a product is remarkable, people will share it.
  • In today’s environment, word of mouth across X, YouTube, Instagram, and communities is a crucial growth engine.

The new rule: think bigger

  • A few years ago, VCs might reject a company for being too ambitious.
  • Now, Anish says, the problem is often the opposite: the idea is too small.
  • Founders should ask:
    • If AI were infinitely intelligent and cheap, how would we reorganize the company?
    • What would the 10x or 1000x version of this product be?

AI, Happiness, and the Human Side of Product

One of the strongest themes in the conversation is that AI shouldn’t just optimize work—it should improve life.

Examples of “loop, make me happier”

  • improve my health
  • help me be a better friend
  • help me laugh more
  • help me feel more connected
  • help me be a more present parent

Why this matters

  • Anish thinks we’ve spent decades building tools that extend the mind, but not the soul.
  • He believes AI can become an emotional/spiritual interface that improves daily life, not just output.

Notable Quotes and Ideas

  • “AI amplifies our agency.”
  • “The loop will help you climb to the local maxima, but then it plateaus.”
  • “More people want to spend time than save time.”
  • “I don’t think it’s a model or capability challenge. It’s a product design challenge.”
  • “Just make more things.”
  • “Building is the new reading.”

Recommended Mindset / Action Items

For product builders

  • Start using AI in real projects, not just demos.
  • Ship something small every week.
  • Ask, before doing any task: How can AI do this for me?
  • Build around joy, not only efficiency.

For founders

  • Think bigger than traditional MVP framing.
  • Don’t overfocus on moats at day one; let them emerge.
  • Reimagine the company assuming AI is cheap, powerful, and widely available.
  • Consider expensive consumer products if they create real value.

For teams

  • Identify where loops can replace manual coordination.
  • Keep humans where judgment, taste, and strategy matter.
  • Train everyone, not just technical staff, to use AI tools.

Lightning Round Highlights

  • Favorite books

    • Conquest and Cultures — Thomas Sowell
    • Seven Powers — Hamilton Helmer
    • Increasing Returns to Scale — Brian Arthur

  • Favorite AI product

    • GrokBot, for being ambitious and practical as a personal agent

  • Best advice / life motto

    • “Don’t discover things through a painful experience that somebody can just tell you.”

  • DJing insight

    • Music creation is becoming more accessible with AI, and he believes music will get bigger as a creative medium because more people can make it, not just consume it.