Over the weekend our team worked on an idea to transform the e-commerce experience as a part of DubHacks 2026. Buying something at the right price takes too much work. You open multiple stores scouring for something and comparing prices, wondering whether waiting and exploring more would save you money. Mercatus offloads the hassle and finds competing offers for the product you’ve already chosen. On a shopping page, click “Wait & save” and choose how long you can wait, and you'll see you're expected savings based on market analysis. The storefront’s listed price becomes the ceiling. A merchant willing to meet your request submits an offer; this is all automated based on discount limits, inventory, and minimum margins set by the merchant, its agent just finds eligible requests. A temporary Visa card authorization backs the purchase request, giving merchants a stronger signal of intent than a product view or an ad click. Merchants compete for qualified demand while controlling what they spend to win each purchase.
One of our biggest learnings was the importance of prioritizing security before shipping a product. We used Vezyl, an AI-powered security auditing tool, to audit our repository for potential vulnerabilities, and our initial scans uncovered 14 critical issues related to the security and safety of our product. Through this process, we learned about prompt injection attacks, how malicious inputs can manipulate AI systems into ignoring their intended instructions, and potential mitigation strategies to prevent unintended behavior. This experience taught us to think beyond functionality and consider how our product could be misused, reinforcing the importance of proactively identifying vulnerabilities and building safeguards into AI-powered applications.
Top comments (0)