Ketch AI

Co-Founded and Architected AI-Driven Retail Recommendation Platform

As CTO and Co-Founder of Ketch AI, I set out with my co-founders to solve a core problem in retail marketing: brands were relying on generic, one-way advertising that consumers increasingly tuned out. We built an AI-driven platform that flipped this model, allowing consumers to opt in to the brands they shop from while our system did the work of identifying sales, scoring them against each user's past purchasing behavior, and surfacing only the most relevant deals. To power this recommendation engine, I designed a data pipeline that ingested historical sales data and used a scoring model to rank deals by predicted relevance to each user's shopping patterns. I architected this as an independent processing pipeline coupled with a shared backend driving business logic across a web app, browser extension, and iOS application — a deliberate choice to meet customers wherever they shopped while keeping engineering headcount and complexity minimal as we worked to prove product-market fit. The results validated the core hypothesis: our platform achieved 66% 30-day user retention, 59% email open rates, and 7% click-through rates (a 10x improvement over the 0.7% industry average for traditional retail email marketing), with users averaging 2 shopping sessions per month. These engagement metrics demonstrated that once users adopted the product, it became genuinely sticky. Despite this strong engagement, we faced a persistent top-of-funnel acquisition challenge. We experimented with referral incentives, paid social campaigns, and distribution partnerships with companies like Capital One Shopping, but customer acquisition costs remained too high relative to lifetime value. Unable to scale acquisition fast enough to match our retention strength, this became a key factor in the decision to wind down the company in late 2025 — a hard-earned lesson in balancing product engagement with sustainable growth economics.