We've been trying out different ways to show product recommendations in the PriceHub.AI chat.
For one of our recent experiments, we picked a fairly ordinary problem: keeping a laptop charged when you're working somewhere without a power socket.
Imagine spending a day by a lake with your laptop and phone hotspot. You're thinking about getting a small EcoFlow, but you're not sure how much battery capacity you actually need.
On a regular shopping website, you'd probably compare a few power stations, check their specifications and try to work out how long each one could keep your laptop running.
So we tried starting with a question instead.
Our first concept shows two compact power stations, along with their battery capacity and weight. But rather than making a final recommendation, the assistant asks for a little more information.

A half-hour walk changes things
What if you have to carry everything to the spot? And your laptop already lasts around five hours on its own?
Suddenly, bringing a 3.5 kg power station doesn't seem so appealing. Especially if you only need a couple of extra hours and can charge your laptop over USB-C.

With that extra information, the assistant moves to high-output USB-C power banks. They store less energy, but they're much easier to carry.
The interesting part is that the second question changes the type of product being recommended, not just the order of the original options.
That's something we'd like to get right in PriceHub.AI.
There are still details to work through. How much explanation should appear alongside a recommendation? When should the assistant ask another question? And how do we fit prices, retailer offers and product specifications into the conversation without overcrowding it?
For now, we're using small examples like this to work through those decisions.



