Built for the problem we kept watching brands struggle with
We're a small team in Seattle. We built Aislegleam because we kept seeing the same thing: smart DTC operators working around a merchandising problem that software was supposed to solve.
Make the shelf space decision for every shopper
The way an expert buyer would, at the speed of software.
A good buyer knows the catalog. They know that the $60 linen throw sells better to a first-time customer than to the repeat buyer who already has two. They know the new arrival in the outdoor section should be surfaced to the segment that's been browsing camping gear for three sessions. They know what goes on the homepage for a Tuesday in November vs. a Thursday in March.
A good buyer can work through maybe 200 SKUs a day. When your catalog is 5,000 or 40,000 products, 199,800 of those products spend most of their time invisible to the right customers.
Aislegleam is the buying judgment operating across the full catalog, per session, continuously. The merchant doesn't need to manage it. They just need to provide the catalog.
The people building this
DYDaniel Yoon
Years building recommender systems at e-commerce infrastructure companies. Saw the catalog-scale merchandising problem firsthand: great models, but nobody had solved the surface delivery and signal pipeline in a way that a practical merchant team could own and measure. Aislegleam is the product he wanted to buy but couldn't find.
PAPriya Anand
ML systems background in ranking and retrieval. Has spent her career working on search and recommendation infrastructure in large catalog environments, where the cold-start problem, session sparsity, and real-time latency constraints are daily engineering constraints, not academic edge cases. Architected the signal processing pipeline and the intent modeling layer.
MHMarcus Holt
Ran marketing and analytics at a growing home goods DTC brand before moving to the product side. Understands the buyer: what problems they're willing to pay for, what proof they need before a pilot, and why integration simplicity matters as much as model accuracy. Leads customer success and the pilot program.
How this started
The specific thing that triggered it was watching a merchandising team at a DTC brand manually re-ranking a homepage carousel every Monday morning. They had a spreadsheet. They'd review the week's sales data, pick the products that seemed to be gaining momentum, and update the homepage order by hand. Six people, a few hours, every week.
The catalog was 14,000 SKUs. The carousel showed 8 products. The math of what was permanently invisible to most visitors wasn't something they'd ever quantified, but the catalog manager understood it intuitively: "Most of our products have never appeared on the homepage. Ever."
That wasn't a content problem or a platform problem. It was a signal processing problem that existing tools hadn't solved at a price point that made sense for a brand at that stage. Enterprise personalization platforms existed, but they assumed a dedicated data team, a six-month onboarding, and a GMV-percentage pricing model that made the ROI math opaque. The merchandising team needed something that connected to their existing Shopify setup, worked immediately, and showed them numbers they could act on.
We started building Aislegleam in 2024. The first pilot ran in early 2025. The model works. The question was always whether we could get it to a merchant team in a form they could actually use.
Say hello
We're a small team and we read every message. If you want to talk about your catalog, ask about the pilot, or just learn more about how the model works, reach out directly.