FAANG Owns the Future of Personalized UI

When my cofounder and I started ATLAS, our thesis was that a small upstart could create the “AI-powered UI of the future”.

We weren’t alone. Since then a number of well-financed startups have launched with the same thesis. Some of the more notable examples: [0], [1], [2], [3]. OpenAI even appears to be exploring it as well.

This is a steelman of the antithesis.

It argues that incumbents with superior distribution — namely: Google and Apple — will almost certainly create whatever personalized UI eventually dominates. The future is theirs to lose.


When we first started out, we thought we were going to create the “AI-powered UI of the future”, or personalized web UI’s. For the first time, AI was going to make it possible for businesses to have bespoke web interfaces, tailor-made to the exact needs of each user. Facebook would look different (have different HTML elements, expose different features, locate features in different places, etc) for the guy who just sells power-tools on Marketplace vs the woman who just wants to see pics of her grandkids.

It turns out that businesses don’t actually want this vision. This is because UI’s serve a double purpose of being how businesses advertise themselves to you. They want to spam you with features you’re not there to use because they want you to discover them. It’s a way of up-selling you.

For this reason, we faced significant headwinds selling this vision to businesses. But we thought that if businesses couldn’t be convinced maybe consumers could be. Consumers — end users — probably have the most to gain by better UI.

But even they couldn’t be bothered to install a chrome plugin to get it. Cluttered UI’s were at worst an annoyance to them. Not even low-vision/blind people really cared. When shortcuts existed for problems they told us they had (e.g. finding a JIRA ticket by its slug) they either didn’t know about them (meaning they didn’t care enough to look for workarounds) or didn’t use them if they did know about them.

There was just no market for personalized UI, even if people genuinely believed it was better. (As a survey we conducted showed they did 70% of the time.)

In hindsight, this shouldn’t have been surprising. People generally have very little interest in paying for UI innovations. Even the iPhone had to be heavily subsidized via 2-year cell phone plans with AT&T.

If people aren’t willing to pay for something, you end up in a tragedy of the commons. No one is going to build it if no one can even be bothered to download it, let alone pay for it. Unless — it can be distributed and monetized another way, e.g. through a service people already “pay” for.

Enter big tech.

They are best positioned to offer personalized UI because they:

Apple and Google can get around the consumer reluctance to install a browser extension because they will be able to add personalization tools directly to the browsers 85% of people already use. And they have already solved monetization for products that no one pays for: by collecting information to run targeted ads.

Any startup that wants to beat Google/Apple to personalized UI will be fighting a losing battle.


Objection 1:

Surely some company will care about having a personalized UI. If we could build it for them and really get the experience right, that could be such a differentiator. Remember: people want this!


Yes, probably someone could be convinced to care. And maybe we could invent some magic UI with AI that differentiates them. But UI innovations aren’t a moat, they are trivial to copy (think: infinite scroll, swipe down to reload, etc). Even if FAANG does not innovate, they will copy what works and give it to everyone for free.

Objection 2:

Browsers and phones are infrastructure. Infrastructure has to be neutral. If Chrome modified websites to tailor them to you — against the wishes of the website — that would be more than a little dystopian. It’s not hard to imagine how bad this would look for Google, a company that people are already pretty paranoid about. (Ask someone whether they think their phone is constantly listening to them.)


You’re right that infrastructure has to be neutral to be valuable. But it can still offer customization options without losing its neutrality. For example, if people can opt in/out of web personalization at whim, or if it allows people to add their own customizations (e.g. Arc allows people to make “boosts”). Apple and Google already make customizations like this possible. See, for instance, iOS shortcuts.