The Market Case for Product Scout

Why I am pivoting from semantic search Interestingly, the process of writing this essay triggered a pivot in my own strategy. While I still believe that semantic search is a vital part of the discovery layer, I’ve realized that focusing exclusively on Shopify merchants is a “slow move” for the broader mission. This does not mean Product Scout is abandoned. Instead, I now see it as a powerful testing ground for the primitives and platform we are building at Poysis.

The Market Case for Product Scout

I built Product Scout on a thesis that predicts how Applied AI will evolve. LLMs will solve the broken discovery layer of the internet, but not in the way most people expect. It will not be a single large platform or a few big players who disrupt industries. It will involve a million domain experts building AI co-pilots across various industries.

The co-pilot thesis is already taking shape. Cursor, one of the biggest successes of the AI era, is a co-pilot for programmers. It is effective because it was built by programmers. The same thing will happen in finance, consulting, food and beverage, and much more. This applies not only to knowledge work, but also to specialized skills and hobbies. Experts will build co-pilots for Excel, as well as for creating a dark academia wardrobe.

It is a lot like hobbyists and bloggers in Web 2.0 building newsletters, email lists, and blogs. Companies like Wix, WordPress, and Webflow came into the picture and enabled experts to focus on writing without worrying about HTML. In the age of Co-Pilots, Product Scout provides the platform and primitives for domain experts to encode their knowledge. Product Scout is to domain expert co-pilots what WordPress was to bloggers. We provide the primitives and interface that help them build, and deploy Co-Pilots.

The first wave of users won’t be everyone. They’ll be the people who already feel the pain of repeating their workflows with AI. The consultant who has a methodology that they repeat across every client. The academic who has a body of knowledge with no distribution layer. The hair consultant who answers the same questions on WhatsApp every sales cycle. These people are already using ChatGPT. They just don’t yet know they can encode their thinking into a product and sell it to others. That’s who we’re talking to first. And interestingly, this does not just power commerce. It opens the door to a new category of internet business. Because currently, domain experts either sell courses or consulting. With Product Scout, they sell AI powered processes.

I have seen people have light bulb moments when I showed them Product Scout. I spoke to the founder of an audio equipment manufacturing company. All I did was show him the search feature, and he started advocating for more features. Features I had documented in the roadmap. Features I did not tell him about. He did not have the language for it, but he was already describing a co-pilot for the people who buy his speakers. He wanted an AI that could help them use their speakers more effectively. And I told him that soon, he would be able to build such things by himself.

This is not theory.

Product Scout is a working name. It reflects what it currently does: power AI search for Shopify merchants. It’s a wedge. In fact, anything else we ship for e-commerce this year is a wedge.

This wedge is necessary for a few reasons:

The market is not yet ready for Co-pilots. One sign of a revolutionary product is that it does not yet exist. But when its intended user sees it, they intuitively know what to do with it. The vast majority of the public did not ask for personal computers, nor spreadsheets. But Apple and Microsoft spent time educating the public, and once it caught on, there was no stopping it. This is the layer we are gunning for within the next two years. People don’t yet know that they need to build Co-Pilots for their hobbies, skills, or industries. But once we show them what is possible, they will.

It serves as a way for us to build trust. If we handle search well, we can slowly add chat, onboarding, and long-term memory. And if we handle all this well, we will have working proof of our primitives powering e-commerce stores.

It allows us to generate revenue within a few weeks of launch.

As such, I am not worried that there are others doing semantic search for e-commerce already. They are not our competition, because we are not playing the same game.

Footnotes

1. This essay started bubbling in my mind on Wednesday. Yesterday, A16z confirmed some of my thesis in their own essay. Read it here.

2. To further understand the difference between a big platform and one like Product Scout, think Salesforce vs Notion. Salesforce builds enterprise-level CRMs. Notion gives regular people the templates and building blocks to build their own CRM without the overhead of Salesforce.