Tech News
← Back to articles

Only three kinds of AI products work

read original related products more articles

The very first LLM-based product, ChatGPT, was just the ability to talk with the model itself: in other words, a pure chatbot. This is still the most popular LLM product by a large margin.

In fact, given the amount of money that’s been invested in the industry, it’s shocking how many “new AI products” are just chatbots. As far as I can tell, there are only three types of AI product that currently work.

Chatbots

For the first couple of years of the AI boom, all LLM products were chatbots. They were branded in a lot of different ways - maybe the LLM knew about your emails, or a company’s helpdesk articles - but the fundamental product was just the ability to talk in natural language to an LLM.

The problem with chatbots is that the best chatbot product is the model itself. Most of the reason users want to talk with an LLM is generic: they want to ask questions, or get advice, or confess their sins, or do any one of a hundred things that have nothing to do with your particular product.

In other words, your users will just use ChatGPT. AI labs have two decisive advantages over you: first, they will always have access to the most cutting-edge models before you do; and second, they can develop their chatbot harness simultaneously with the model itself (like how Anthropic specifically trains their models to be used in Claude Code, or OpenAI trains their models to be used in Codex).

Explicit roleplay

One way your chatbot product can beat ChatGPT is by doing what OpenAI won’t do: for instance, happily roleplaying an AI boyfriend or generating pornography. There is currently a very lucrative niche of products like this, which typically rely on less-capable but less-restrictive open-source models.

These products have the problems I discussed above. But it doesn’t matter that their chatbots are less capable than ChatGPT or Claude: if you’re in the market for sexually explicit AI roleplay, and ChatGPT and Claude won’t do it, you’re going to take what you can get.

I think there are serious ethical problems with this kind of product. But even practically speaking, this is a segment of the industry likely to be eaten alive by the big AI labs, as they become more comfortable pushing the boundaries of adult content. Grok Companions is already going down this pathway, and Sam Altman has said that OpenAI models will be more open to generating adult content in the future.

... continue reading