Why Guidance, Info-Seeking, and Writing Apps Are Features, Not Companies

OpenAI released a report on how people are using ChatGPT. Here’s my take:

Investors chasing the current wave of AI apps that give you practical guidance, fetch information, or write your emails are buying into a familiar illusion: short-term traction equals long-term value. It doesn’t. These categories are destined to be short-term wins and long-term losses, and the reason is simple—platforms always win.

Look at who sits atop the AI stack today: OpenAI, Anthropic, Google, Apple. They don’t need to “build apps.” They own the foundation—the models, the ecosystems, and the distribution channels. And history shows us what happens when platforms control the foundation: standalone apps in horizontal categories become features overnight.

Remember the 1990s search wars? AltaVista, Lycos, Ask Jeeves—all hot IPO stories until Google showed up and redefined the category. Same with email: once a dozen providers offered it, now it’s Gmail, Outlook, and Apple Mail, all bundled and free. Blogging platforms? LiveJournal and TypePad faded once Facebook and WordPress swallowed the market.

The same playbook is unfolding with AI. Standalone writing apps? Microsoft has Copilot in Word and Outlook. Guidance apps? Google is already weaving Gemini into Workspace and Android. Information-seeking bots? Apple’s Siri—laughable a decade ago—will be rebooted as a system-level AI that lives in your iPhone. And OpenAI’s ChatGPT is moving from “app” to “operating layer,” embedding itself into everything from browsers to phones.

For a small AI startup, that’s checkmate. You can grow users fast, sure. But your “moat” is shallow. The platforms have the distribution, the data, and the user lock-in. They don’t need to buy you—they can just outlast you and roll your core feature into their ecosystem.

And then there’s regulation. An AI app giving legal or medical “guidance” might be fun until regulators step in. History is littered with examples: early health forums, peer-to-peer lending startups, even ride-sharing before city councils got involved. Compliance takes deep pockets, and only the big platforms have them.

So yes, investors will see adoption spikes. The apps will demo well. They’ll even raise money. But over time? They’ll suffer the fate of Pets.com, Groupon, or flashlight apps—fun while they lasted, irrelevant once the platforms swallowed their functionality.

If you’re betting on AI, bet on the verticals with workflow integration, regulatory defensibility, and domain-specific data. Everything else? Features masquerading as companies.