If you have been doing online shopping lately, you know how hard it is to buy that one product that is actually built for your use case.
It might take you days to finally get to that one product as you keep researching. The reviews, influencer suggestions, and family and friends’ opinions also add to the time spent.
So what if you could add a layer of AI that helps you make that choice using the same plain text you already use while searching? For example, instead of jumping between tabs, you could simply ask something like, “Give me a list of mobile phones for my mom that are under $200, easy to use, and decent enough for photography.”
Hi, I am Divanshu. I manage marketing at Scrapingdog. I built this product recommendation app using Lovable along with Scrapingdog’s Google Shopping API and Amazon Search API. It gives you product recommendations based on a single prompt.
This app reduces the hassle of opening 20 tabs just to find the best fit. It saves you time and keeps your system from slowing down in the process.
In the end, I will share with you the public link of the app too, so that you can try it out. To access, you would need the two API keys, one for Scrapingdog, the other for Openrouter (it is an aggregator for AI models, we are using OpenAI in the process).
How Does This App Work
- User need to enter the API keys for Scrapingdog & Openrouter.
- They can write their prompt in simple English, describing their specific requirement.
- The AI layer in the app understands the request, optimizes it, and then gives it to the Google Shopping API & Amazon Search API.
- The results are then shown at the frontend with a card for each product & link.
- Users have the option to land directly to product page using the link given at the bottom of the card.
How Did I Build This App
The idea came from my own research, where I was looking for a product, found myself kept on searching for endless options out there & at the very end was not able to find one.
So basically, I created the prompt for lovable using ChatGPT & then did iterations on it to make it pinpoint perfect.
Tips To Use This Web App Better
First and foremost, the prompts should be clear & concise, like “give me a list of laptops that would cost under $500, good for productivity”.
An open prompt like “Best Laptop” won’t give the desired results from this App.
You can also clearly mention the use case. Saying who the product is for and how it will be used helps the app narrow things down better.
Adding a rough budget always helps. Even if it is not exact, a range like under $300 or around $1000 makes the results more relevant.
Try not to mix too many categories in one prompt. Requesting a phone and its accessories together may confuse the results and make them less accurate.
If the first result set does not feel right, tweak the prompt slightly instead of starting over. Small changes usually improve the output.
How To Use This App
👉 Click here to launch the web app. ⬇️
You need to register to openrouter, which is an aggregator of AI models. If you are already using ChatGPT’s API, you need to change the backend where the endpoint from open router is used.
The other API that will be used is from Scrapingdog. You get 1000 free credits to test this API, here is the link to register & claim your credits.
The first screen that you will land on when opening the app through this link will be: –
When you have both of your API keys, you can enter them here and click continue.
Eventually, you will land on this screen, and you can start prompting here.
And here is the video of how this app works:-
Conclusion
Since we are using e-commerce APIs, you may build your own web app like this using Lovable and APIs from Scrapingdog.
Now we offer other e-commerce APIs as well that you can check out and build more of such apps for your use cases.
Recently, I built a blog where you can track your listings search rankings on eBay. Do check that blog too here.
Additional Resources
With that being done, we do have multiple tutorials for non-coders to build their own tools. I am linking to some of them that we have already built & you can use them as a reference.
