a16z Podcast: Consumer AI's Real Winners Are Personal Agents, Not Bigger Models

SPEAKER_1 (0:01)

So we've been tracking the Top 100 Consumer AI Apps, and the seventh edition just dropped. What's jumping out at you this time? Honestly, it's not about bigger models anymore. The real story is personal agents, changing behavior, and the first signs of actual durable businesses emerging. And here's a striking stat: about half of Americans now report using AI, but only around 4 percent pay for subscriptions. That's a massive gap. Though I've heard the top 1 percent of users are spending heavily, right? Exactly. It feels like a power-user market more than mass-market. And here's what's wild—many of the top products by spending don't even show up on traffic charts. Revenue and usage are completely diverging. So what are most consumers actually doing with AI right now?

Still pretty basic stuff. Replacement for search, homework help, writing emails. The strongest categories are focused on "getting things done"—productivity, editing, photo and video design, search. But the bigger opportunities might be elsewhere. Like where?

Networked categories. Dating and recruiting, for example. AI hasn't produced a breakout product there yet. Shopping, home buying, retail, gaming, entertainment—all wide open. Let's talk monetization. What's working?

Most revenue today comes from subscriptions, which is actually unusual compared with the old consumer internet. That may change as inference costs fall and ad products mature.

OpenAI's ad rollout is fascinating—they've already hit a billion-dollar annual run rate through advertising. But there's a trust issue with ads in AI, isn't there? Absolutely. Ads must be carefully labeled and introduced gently, or they risk breaking trust. It's a delicate balance. What about the competitive landscape? Is ChatGPT still dominating?

ChatGPT remains dominant, but movement's happening underneath. Claude just passed Gemini in paid subscribers. And specialized tools like Suno, 11 Labs, Midjourney, Granola, Whisper Flow, and Superhuman are showing how focused products can win by serving specific jobs better than broad platforms. So these aren't just wrappers around ChatGPT?

Not at all. They're rich software experiences built around personal context. That's the key distinction. Where do you think this is all heading?

Right now, AI mostly helps people save time. But the real breakout companies may be the ones that help people spend time— create, tell stories.

We're in a world where we got excited about ChatGPT turning things into a conversation. And the next phase? Agents and software that turn that conversation into action.

The market is still early, still concentrated, and still full of white space. That's what makes it exciting.