Emin Gölemen

At Reiterate, I've worked from early problem framing and concept exploration through design, iteration, and launch on an agentic workflows product for accountants. The product lets users automate repetitive accounting tasks through a chat interface, connected to their accounting software.

The hardest product design problem was not how to make the AI feel impressive. It was how to make its work trustworthy.

Accountants did not want an AI system that took data in and produced a result without explanation. They needed to understand what the AI was doing at every step. They wanted to see decisions, inspect reasoning, verify source data, and stay in control before anything was posted into accounting software.

That meant the interface could not be a black box. The workflow needed to make the AI's process visible, especially when it was making decisions from financial data.

To support that, we designed the workflow as a visible canvas. Every step is connected and inspectable. Each step has its own conversation, output, reasoning fields, and confidence rating, so users can audit the result and trace it back to the beginning. Users can open the original invoice files with one click to visually confirm the data.

There is always a final approval checkpoint before posting into accounting software. The AI can also request approval mid-process when its confidence is low. If something goes wrong, users can restart from the failed step instead of rerunning the whole workflow, saving time, cost, and trust. The result was not just automation, but automation accountants could inspect, correct, and approve with confidence.

This work shaped how I think about AI product design. A chat interface is not just another input pattern. In an AI product, the designer has to think about delegation, uncertainty, recovery, confidence, and control. The experience has to help users understand not only what happened, but why it happened and what they can do next.

At Hub88, part of Yolo Group, I was the only designer for HubConnect, a B2B platform serving game suppliers, iGaming operators, and internal operations teams. I worked with a team of 20+ developers and a wide range of internal and external stakeholders. Because I was the only designer on HubConnect, design work often became the shared language between legal, operations, product, and engineering.

This was my first deep B2B product role, and it changed how I thought about product design. The interfaces were data-heavy, the workflows were operationally sensitive, and the stakes were high. In iGaming, a confusing setting or missing piece of information can create real business risk.

This became especially clear in due diligence. As Hub88 grew, both the number of companies using the platform and revenue increased roughly 4x. Several companies told us they were switching from other tools because our back office platform was better than the alternatives. That growth was a strong signal, but it also put pressure on the operational systems behind onboarding.

More operators wanted access to games as quickly as possible. But due diligence still depended on long email threads, PDF forms, and manual back-and-forth with the legal team. As demand grew, that process became harder to manage and harder to audit.

The needs were different on each side. Operators cared about speed. Suppliers wanted more distribution, but only through operators who had passed a safe and auditable due diligence process. Hub88's legal team needed visibility, structure, and a way to manage growing workload without losing control.

We designed a due diligence flow inside HubConnect where operators could self-onboard through a structured step-by-step form. Legal could manage active reviews from a task board, track what was missing, and keep the process auditable. For suppliers, the process gave confidence that access to their games was expanding safely.

At n11.com, I worked on a mobile-first marketplace used by around 30 million unique shoppers each month.

The team was highly cross-functional. Each project involved UX researchers, UX designers, UI designers, optimization specialists, UX writers, product owners, and engineers. Design decisions were shaped by both user behavior and experiment results, not just interface preference. It was an environment where craft, research, experimentation, business goals, and operational constraints all had to meet in the shipped experience.

One of the strongest examples was UçUç Kupon, a personalized loyalty and coupon experience. The design challenge was not only to make coupons easy to find and use, but to make the whole incentive system understandable: how shoppers earn value, what they can use, where it applies, and why they should come back.

I designed across the full mobile experience, including discovery surfaces, the coupon wallet, points history, coupon creation, sending points to others, checkout redemption, category personalization, and empty/error states.

The product had to make a complex incentive system feel simple at shopper speed. Users needed to understand what they had earned, what it was worth, where it applied, and how to use it without slowing down the purchase journey.

This work, together with other major optimization initiatives such as the product detail page revamp, contributed to significant conversion improvements. The larger initiatives I worked on showed around a 25% relative lift from a roughly 2% conversion baseline.

I also effectively operated as a DesignOps lead while at n11. I helped create the design system from the ground up, supported the team's migration from Sketch to Figma, and maintained the system as new product needs emerged. That work made design quality less dependent on individual effort and more supported by repeatable systems, shared habits, and clearer team workflows.