Rolle AI

Rolle AI is a second-hand electronics marketplace that pulls listings from local resale platforms into one place, then uses AI to do the work buyers normally do themselves: filtering out junk, comparing prices across the full market, and negotiating with sellers on your behalf.

The platform is built around four pillars: Aggregation, Verification, Negotiation, and Warranty. Together they turn a fragmented and stressful buying process into a guided, trustworthy experience where the user is always informed and always in control.

The second-hand electronics market is fragmented and opaque. Buyers jump between platforms, compare listings with inconsistent condition descriptions, and have to negotiate directly with strangers, with no real way of knowing whether a price is fair or a device is what it claims to be.

Rolle's pivot was to take ownership of that entire process. The design challenge was to surface a lot of real-time data (prices, seller quality, negotiation activity, deal status) without overwhelming users, and to keep them feeling in control at every step, including when an AI agent was acting on their behalf.

I designed the platform around four trust layers.

Aggregation pulled listings from local resale marketplaces into a single searchable feed. Every product card in search results and on the homepage included a condensed version of the price spectrum: a color-coded bar showing where a listing sits across the full market. The full version lived on the product detail page and mapped five reference points: risky (dangerously low prices that signal a suspect listing), used, refurbished, new, and the fair price Rolle calculated from the aggregated data. Buyers could see at a glance whether a deal was reasonable or a warning sign.

Verification filtered that aggregated feed through an AI model that read each listing's description, photos, seller rating, price, and other metadata, then removed scams and low-quality results before users ever saw them. On the product detail page, buyers saw the full price analysis alongside every top verified listing, with the option to visit any listing directly or ask Rolle to negotiate on their behalf.

Negotiation removed the most stressful part of second-hand buying. An AI agent contacted all relevant sellers simultaneously and worked toward the lowest price. I designed a live activity log showing how many sellers had been contacted and the stage of each conversation (for example: contacted 4 sellers, 2 replied) so the process never felt like a black box. Once a deal was secured, users chose their path: go to the vendor's marketplace and finalize the deal themselves, or let Rolle handle the checkout and include warranty automatically (the recommended option).

Warranty was a complete flow from activation through to checkout. Rolle offered one year of warranty and damage protection on every device bought through the platform. I designed the full activation flow, confirmation modals, and checkout experience for users who chose the Rolle-managed path.

My central challenge across all of this was making complexity feel manageable. The platform generated a lot of data and activity that users genuinely needed, but presenting it without care would have felt overwhelming rather than empowering.

Two pieces of work shaped the experience most.

The negotiation timeline was the spine of the post-deal experience. I designed it to cover both completion paths with the same milestone structure: negotiation started, negotiating with sellers (with the live activity log), deal secured, and then a clear fork into the two flows. Each state gave users a progress update, a choice where it mattered, and explicit handoff instructions so they always knew what was happening and what came next.

Gem Alerts let users mark any device as a gem and get notified the moment it appeared at the best condition and price in the market. It was a low-effort way to stay in the loop without having to check back manually, and it reinforced the feeling that the platform was actively working for the user in the background.