While building Lima AI and watching our AWS bills climb thanks to OpenAI's APIs, I've been having some existential thoughts about defensibility. You know, the kind of 3 AM reflections that make you question whether your startup will survive the AI revolution or become another cautionary tale in some future entrepreneur's Medium post.
The big question haunting every founder right now: Is AI going to disrupt every business? And if so, how can we predict what survives the digital apocalypse?
For lack of better crystal balls (and because I'm apparently a masochist who enjoys overthinking), I've developed my own delightfully simple mental model for AGI, slightly inspired by Ted Chiang's brilliant storytelling:
"AGI will be an entity—a voice—that knows everything and can answer any question. No more, no less."
This means AGI will know how to build anything ever invented by humans, broken down into simple, actionable steps. But here's the catch: it still faces two fundamental constraints—law and physics.
I won't be able to build a nuclear bomb without uranium (thank goodness), but I will be able to build a YouTube competitor. However, I still won't be able to magically convince people to abandon their favorite cat video platform for mine. Human behavior, it turns out, is harder to hack than code.
This got me thinking about Hamilton Helmer's excellent book "7 Powers: The Foundations of Business Strategy." Helmer identifies seven sources of enduring competitive advantage. But in an AGI world, which of these powers will crumble like a house of cards, and which will stand firm like that one Nokia phone from 2003?
Let's dive in:
What it is: The cost advantages that come with size—think Amazon's impossibly efficient fulfillment centers that would make Santa's workshop weep with envy.
AGI's impact: Can AGI magically give me Amazon-scale infrastructure without Amazon-scale money? Sadly, no. AGI might optimize my operations, but it can't conjure up billions of dollars or teleport warehouses into existence. Physics and capital requirements remain stubbornly real.
Verdict: Scale economies remain a fortress that AGI can't easily breach.
What it is: The beautiful phenomenon where your product becomes more valuable as more people use it—like how YouTube creators go where the audience is, and audiences go where the creators are.
AGI's impact: Can AGI disrupt network effects? It's like asking if a really smart person can convince everyone at a party to move to an empty room. Sure, they might have great arguments, but people tend to stay where the action is. Network effects create gravitational pulls that even superintelligence struggles against.
Verdict: Network effects remain as sticky as ever.
What it is: Positioning your business to exploit competitors' weaknesses or blind spots—finding that sweet spot they can't or won't address.
AGI's impact: Here's where things get spicy. If AGI can help me identify market weaknesses, it can help my competitors too. It's like everyone suddenly getting the same cheat codes. The advantage of superior market analysis evaporates when everyone has access to the same superintelligent consultant.
Verdict: First casualty of the AGI revolution.
What it is: Making it painful (financially or operationally) for customers to leave—like how switching from iPhone to Android feels like learning a new language.
AGI's impact: AGI could theoretically help customers navigate complex transitions, acting like the world's most patient tech support. But here's the thing: many switching costs are contractual (hello, 3-year enterprise deals) or emotional (my entire photo library is in iCloud). AGI can't rewrite contracts or emotional attachments... yet.
Verdict: Survives, but with some battle scars.
What it is: The premium customers pay for trust, status, and emotional connection—why people buy Ferrari instead of "equally fast car X."
AGI's impact: This one hurts. Ferrari's brand power comes from decades of racing heritage and Italian craftsmanship mystique. But AGI can help anyone build an objectively superior car and craft a compelling brand story. When everyone has access to perfect marketing strategies and flawless execution, brand differentiation becomes much harder to maintain.
Verdict: Brands will need to work much harder to justify their premiums.
What it is: Exclusive access to scarce, valuable resources—oil reserves, prime real estate, or that one engineer who actually understands your legacy codebase.
AGI's impact: AGI might help me find new oil reserves or optimize resource extraction, but it can't change ownership. If you own the lithium mine, you still own the lithium mine. AGI might shift which resources are valuable, but scarcity and ownership remain fundamental economic realities.
Verdict: Resource owners sleep well at night.
What it is: Superior operational processes, innovative production methods, and management systems that consistently outperform competitors.
AGI's impact: Process power is essentially institutionalized knowledge—and knowledge is exactly what AGI democratizes. Those secret sauce operational efficiencies? AGI knows them all and shares them freely. It's like having the world's best consultant available to everyone for the price of an API call.
Verdict: Process advantages become commoditized overnight.
🏆 WINNERS: Scale Economies, Network Economies, Cornered Resources
💀 LOSERS: Counterpositioning, Branding, Process Power
🤔 MAYBE: Switching Costs
At Lima, we're building tools for community creation. Looking at this analysis, it feels like we're betting on the right horse—network economies are AGI-resistant, and communities are fundamentally about human connections that can't be easily replicated by algorithms.
But here's the real insight: the future belongs to businesses that can combine AGI's capabilities with inherently human or physically constrained advantages. The winners will be those who use AGI to enhance their scale, networks, and resource control—not those who rely on AGI-replaceable knowledge or processes.
So while AGI might make us all smarter, it won't make the fundamental laws of business physics disappear. Some moats will be filled in, but others will only get deeper.
Now, if you'll excuse me, I need to go optimize our community-building algorithms while they're still defensible.
2026-02-10