The Digital Factory – One of the elements that differentiates Perplexity AI from Google is the simplicity of the design. Why did you make this choice?
Henry Modisett – What users want is text and images. It's not an interface, it's not a brand. People just want to read. They want to read in an enjoyable way. So a lot of our focus in the beginning was just showing the right content. We then added other elements, like colors, typography, animations. But always in a way that didn't ruin the reading experience. We change the design every week. We're always adding new features or tweaking something that already exists.
Advertisement
We launch new ideas about as quickly as we did when we started. The challenge is not to ruin the original experience. It’s about improving it, learning how people use Perplexity and trying to address use cases. For example, we noticed that a lot of people were searching for places, so we integrated a map, but only when it’s useful.
At first glance, Perplexity seems like a chatbot like any other. How do you manage to make your difference understood to someone who is coming for the first time?
As a free product for consumers we only have 300 milliseconds of attention span. Think about how they got there. Some clicked on a link, others just heard the name. Our goal is to push them to do something very quickly. The worst method would be to try to explain what we are. The best method is to get them to try it to give them an idea of what Perplexity is and what it can do. That's why we offer suggested questions on the home page. This lowers the barrier: you can try without even having to type anything.
Your approach is very different from Google's, which has to integrate AI into a product that has made it successful. Does this give you an advantage?
Advertisement
We're inventing something that didn't exist before, by making it possible to ask questions and get answers quickly and reliably. It's very difficult to take an existing product and evolve it in a different direction. In reality, our product is like Google, but in reverse. Google was designed to help you navigate the web, by sending you to websites. Whereas our product is about taking information and presenting it to you without you having to go anywhere. It's very difficult to want to do both at the same time (as Google does, editor's note). Unless you can guess exactly when the user wants one or the other, it makes the product very unpredictable.
You start adding widgets to Perplexity. For what ?
A lot of information isn't really well represented by just text. If you ask for the weather, that's not the best way to see what the weather is like. We started showing a little weather widget. One of the things we noticed is that people are asking questions that AI isn't helpful for. For example, people are typing in calculations. That's a waste of resources. We can just program a calculator. We're trying to figure out when AI is really helpful and when it's not. In the case of weather, it's not helpful so we just use an API.
You announced the upcoming arrival of advertising on Perplexity. What will this look like?
It’s not defined yet. But we don’t want to ruin the user experience that makes Perplexity great. It would ruin everything we do. A lot of searches are driven by the desire to buy something. And AI has proven itself to be good at that kind of search. I think we’re going to find a way to add ads, in a very natural and really useful way.
But Perplexity was designed so that you never have to click on links. But if you have ads, does that change the way you have to think about the product?
Ads work two ways today. They're either based on click performance or based on impressions. But those are concepts that Google and Facebook created, it doesn't have to work that way. The only thing that really matters is that companies are making money off of Perplexity. So I don't think we have to follow the way advertising works today.
Selected for you