The oft-discussed enshittification thesis has pervaded modern life in every respect. Everything involving a screen somehow feels like a ripoff, and everything seems to get worse online overtime. Freemium services get monetized, surcharges and service feeds accumulate, more and more stuff gets paywalled. This is well-documented via how many times you simply have to click “log-in to read article” or “accept all cookies” or “personalize your ad feed” or whatever. Before the advent of targeted advertising you used to just exist and be fed products; now you have the honor of selecting how you are preyed upon.
Shrinkflation is the other real-world side of this. Products deliver less value while remaining the same price. I wouldn’t chalk this up to shareholder greed or anything, I would honestly just call it lack of innovation in value delivery combined with natural competitive margin erosion. Larry Hamtil had a great tweet (referencing a piece I wrote on Apple/Ubiquiti) regarding product development and value delivery:
The point I’d like to make is that AI will now be a tool to rationalize delivering shittier products and services at the same price points. Necessarily, in order to justify certain valuations, AI/LLMs must take margin from both B2C and B2B businesses. Whether that means shittier vibe-coded apps or AI agents pinging homeowners to make shitty low-ball offers, or simply trolling e-commerce sites for deals all day, things will probably get worse before they get better. And prices will stay the same, meaning it will open up undercutting opportunities for specialists in the field. You can (and should) compete on price in a world where AI agent competitors (who are all getting their information from the same data feeds) are bidding against you.
De novo innovative product differentiation will confidently remain a way to make money. Maybe an LLM could have dreamed up Nespresso or Ubiquiti APs on its own, but it probably would have concluded that “no the upmarket differentiation opportunity in those markets is limited and competing on cost is doomed”. I wouldn’t blame them—I certainly would have thought that!
One example of humans beating out robots via differentiation is actual live-human travel agents, who are making a comeback. It’s truly an awful experience to try to book your own vacation and confirm everything via the internet these days. Farming it out to an actual real-life human is an actual value-add. A person who can talk to the other humans actually working at the airline or accommodation provider has become so valuable that the career has been revitalized even as AI agents book travel and online travel agencies grow their businesses. “Headache Solver” or “I’ll Just Do It For You And Send You An Email With Everything” is another name for these people, and I think these service providers will continue to be insulated from disruption despite what AI-maxis will claim. I myself simply don’t want to interact with a robot, even if it is cheaper, and will pay more to have someone fix my problems (even if that someone is just using a robot).
I haven’t thought of anything, but to any entrepreneurs out there I’d give the following advice: think of a service or product that was better 5 years ago. Develop a premium version of it. Do it at a reasonable markup and deliver it consistently and no one will blink an eye. Nespresso charges more than your local Greek diner, Ubiquiti charges more than T-Link; people don’t care. You can differentiate where you feel it might be the most unlikely. Particularly as a new wave of AI-driven enshittification sweeps across both the digital and physical worlds.
Good stuff.
No one is coming to save you.
Nobody owes you anything.
Things are the way they are, find a way through. (...is what I keep telling my teenage kids)
What a time to be alive!
...or AI can simply charge much more in the niches where it excels and is delivering greatest roi.
friends at a private and mid-size tech companies uniformly state they would pay 10-100X above their high tier token buffet. they are def not using AI for searches and gif memes.
weak but reasonable analogy; hyper-essential tsmc allowed >90% of the sector profits to go to apple, amd, ....
not just for a few years, but decades.