Key Takeaways

  • Indie hacker Pieter Levels pivoted to AI applications in 2022 and now claims pure software features offer zero long-term defense against competitors.
  • According to Levels, only six defensible moats survive AI commoditization: proprietary data, network effects, audience distribution, physical infrastructure, capital, and regulatory capture.
  • Sam Parr admitted his own stance shifted within a month: “3 weeks ago I was like it can't be that bad. It might be. It might be that bad.”
  • Marketplace dynamics still trump product quality: Parr highlights that platforms like Craigslist and eBay retain users despite terrible interfaces because everyone is already there.
  • Shaan Puri argues founders should build where AI cannot replace the core offer: run physical operations or community-driven businesses, then use AI to slash operating overhead.

The Disagreement

Ben Levy brought Pieter Levels' viral warning to the table. In 2022, Levels shifted his focus entirely toward AI apps. Now, Levels warns that traditional software-as-a-service is facing an extinction-level squeeze. Anyone with an LLM prompt can replicate a workflow wrapper over a weekend. If your business is merely software that converts input A to output B in a browser, your moat is gone.

Parr finds himself agreeing more than he expected. While he wondered if Levels was sounding an alarm prematurely, Parr conceded that software commoditization is arriving faster than founders want to admit. “If you look at some of the greatest companies of all time, their products might totally suck,” Parr noted. “Craigslist, eBay, a bunch of different marketplaces fall into this, but you use them because that's where everyone else is. And so I think that network effects is probably one of the most important.”

Puri sees the shift differently. He doesn't believe founders should abandon building things just because pure software code is cheap. Instead, he argues founders should invert the target. “I think you want to be doing things where AI doesn't do the core of what you do, like fundamentally can't do the core of what you do, but it makes what you do easier,” Puri explained. For Puri, the solution is pairing cheap software labor with physical operations, local services, or in-person communities that code alone cannot duplicate.

Who's Right (and When They're Wrong)

Levels is right about single-utility software tools. If you run a micro-SaaS that charges twenty dollars a month to format PDFs, transcribe audio, or generate landing page copy, you are running on borrowed time. When code generation drops the cost of building software to near zero, features stop being assets. They become commodities.

Where Levels overstates the case is assuming every buyer wants to build or re-platform their stack. Software adoption is driven by inertia, procurement approvals, compliance, and trust. Enterprise buyers do not switch vendors just because an indie hacker spun up an alternative on a Friday night. Incumbent distribution and workflow lock-in survive longer than pure tech arguments predict.

Puri's model is right for new founders starting today with zero users. If you do not have an existing distribution channel, do not launch another standalone dashboard. Build a business anchored in the physical world, whether that is local services, manufacturing, or events. Use automated tools behind the scenes to run the back office at tenth-of-the-cost margins while your competitors drown in manual payroll.

What to Do With This

Audit your current product backlog by asking one question: if an intern with an AI coding assistant cloned our top three features by next Tuesday, would our customers actually leave? If the answer is yes, you do not have a software business; you have a temporary feature. Shift your roadmap immediately toward acquiring proprietary data from your users or building two-sided network loops where users bring each other.