Inspiration

I love deals and I love quality food. Those two rarely go together, but when they do, I stock up. The problem is catching those moments. Sales on the brands I actually want are scattered across weekly ads, digital coupons, and cashback apps, and by the time I notice, they're usually gone. So I built a tool that finds those rare intersections for me.

What it does

Promobeaux lets you build a watchlist of your favorite brands and items, then choose the stores you actually shop at. It searches for current sales, coupons, and cashback offers on those exact items and alerts you when one is worth acting on. Each deal is compared against the price you usually pay, so you only see real savings, not noise. Your deals are grouped by store, with a running total of how much you could save this week.

How we built it

I built Promobeaux in Emergent, using its AI agents to generate and iterate on the full stack. Deal discovery runs on Exa, which searches the web for weekly ads, digital coupons, and cashback offer pages for each watchlist item at the user's chosen stores. A scoring layer then compares each result against the user's usual price and only flags it as a real deal if it beats that price by at least 10% or includes a meaningful promo like a BOGO or digital coupon. I added a "Check deals now" button for on-demand searches alongside a scheduled daily check.

Challenges we ran into

Real grocery price data is hard to get. Most stores and cashback apps don't offer public APIs, and connecting to users' store accounts raised privacy and security concerns, so I chose web search instead. That created a new challenge: search results are messy. Filtering out expired promos, irrelevant products, and pages without clear prices took real work. Defining what counts as a "deal" was its own problem too, since a sale price means nothing without knowing what the shopper normally pays.

Accomplishments that we're proud of

I went from idea to a working, polished app in about a day. I'm proud that Promobeaux judges deals against your own usual price instead of trusting a store's "sale" label, which is what makes the alerts worth paying attention to. And I built something I'll genuinely use every week.

What we learned

Data access shapes the product more than features do. Knowing where your data will come from before you design anything saves hours. I also learned how far an agentic builder can take you when you give it a clear user, a tight scope, and specific rules for what "good" looks like.

What's next for Promobeaux

  • Launching on Whop so other deal lovers can use it
  • "Save the difference": moving the money you saved on a deal into a savings goal
  • Learning each user's usual prices automatically from the deals they act on
  • Barcode scanning to add items in seconds while in the aisle
  • Quality filters, so you can prioritize organic, clean ingredient, or favorite brands
  • Direct store data where official APIs exist, like Kroger

Built With

Share this project:

Updates

Submission history