Perplexity, the AI-powered search engine backed by Jeff Bezos, Tobi Lütke, and other notable investors, has just launched a new feature called Buy with Pro. This AI-assisted shopping search engine will let US-based Perplexity Pro subscribers purchase products without leaving the AI search engine. Unlike traditional search engines such as Google, returning a list of ranked sources you must read through to get the answer to your query, Perplexity’s AI search instead reads those ranked sources and summarizes the answer.
The AI Commerce and Shopping Search Experience
The AI commerce experience aims to add a layer on top of AI search that helps users find products that meet their criteria using conversational messages instead of keywords you would type in Google. Users interact with Perplexity in natural language, asking questions and the “prompts,” and then Perplexity will understand that query and return search results, including product recommendations. The new Buy with Pro button allows users to check out using one click on Perplexity’s website. In addition, they have a nice feature called Snap to Shop, which is essentially a visual search like Google Lens. You take a picture of an item, and Snap to Shop can find similar products.
The Old vs. The New
With a traditional Google Shopping Search, the Shopper enters some keywords into a very old-school interface, which generates a list of product pages that take them to each retailer’s website. After browsing through the images, users must click on each product image that appeals to them and read product specs, read lots of reviews, and figure out whether it’s the right fit or at least a contender.
Whereas what Perplexity is trying to do is make things more intelligent for the Shopper by returning not just a list of product pages but a list of what they call product cards, which summarizes all the information for you with statements like “Buy this if you want a luxurious, oversized cashmere sweater that combines comfort with a timeless, stylish design.” Product cards include high-resolution photos, a summary of the reviews with sources, key features, etc. These are within the Perplexity conversation and in a straightforward format that the Shopper can digest.
Also helpful…users can ask follow-up questions and/or refine the results by updating the conversation, etc. For example, add guidelines such as color, flavor, size, or price.
Our Real Test for an Intended Product
We started our test with a very simple query: “Looking for a cashmere sweater for a party to go with brown pants.” The first results were all the women’s pants!! We didn’t ask for pants; we asked for a sweater to go with pants. And none of the results had the option to Buy with Pro, only to visit the merchant’s website.
We started a new interaction with the identical prompt. This time, we received brown sweater suggestions, which is closer (do people wear brown sweaters with brown pants? maybe…), but still all women’s styles, while our user profile is configured as a male. So we should be receiving suggestions for men’s sweaters. Again, there were no Buy with Pro options, but we did find a Buy with Shop Pay option this time. Clicking the button brings the user to the Seller’s Shopify store directly to the checkout page to complete the purchase outside the Perplexity environment.
On the third attempt, we appended the word ‘mens’ to the end of the prompt. This time, the results were mostly what we would expect, besides disagreement about what color sweaters would go with brown pants, though there was still no Buy with Pro option.
Perplexity Missed the Mark
While this use case for GenAI seems to be a natural fit for GenAI technology, it’s missing the mark. Search results were poor in our tests, suggesting product-market fit might be a long way away. The current solution lacks enough data or the needed intelligence or context.
We’re optimistic about the future of this technology, but in our limited trial/quick test, we weren’t impressed. Generative AI should be able to understand our profile and what we’re looking for and return a men’s sweater and not pants.
It Ain’t All Bad
While search results for the sweater didn’t deliver against expectations, the results did include some thoughtful suggestions about shades and tones to consider for different looks that the shopper might be interested in achieving, as well as tips for accessorizing the outfit:
So, there is promise, but is the product ready? We don’t think so. Again, it’s early, and this was our first couple of queries. But for paid Perplexity Pro users, that’s their first impression, and Shoppers will be disappointed, and Shoppers are very discerning.
Final Thoughts on Perplexity
Perplexity is still a young startup and may figure it out. But trying to compete with the OpenAI’s of the world and launching products prematurely may be a bad idea.
Whether expanding from the Shopify partnership to increase the scale of Buy with Pro or working on understanding the queries better and returning better results, the next move should be figuring out and bridging the gaps. In the era of AI, capturing interest and early users for only $20/month is the easy part. Keeping them, not so much. You can also listen to a recent Perpelixty AI Commerce Search podcast here .