Over a two-year period, a $35M+ ARR Instructure product was transformed into an AI-native solution and deployed at scale for more than 2 million users worldwide — introducing explainable and compliant AI into the learning ecosystem without compromising trust, accessibility, or operational stability.
AI-Based Prediction Market
Depot
I was focusing on revamping Depot's comprehensive CI/CD developer experience, covering everything from pipeline creation and secrets management to diagnostics, logs, artifacts, and governance. The goal was to minimize friction in essential processes, simplify intricate steps, and provide an accessible, scalable design system that could grow with the product.
AI-Based Prediction Market
Raiffeisen
Over 2 years the full mobile banking ecosystem was rebuilt and deployed across 16 countries, resulting in a +670% increase in app downloads and more than €300M in additional annual digital transaction volume — driven by restored trust, a new design system, and a redesigned end-to-end experience.
AI-Based Prediction Market
Bitpanda
Bitpanda is the trading platform that Raiffeisen Bank customers use when they buy and sell digital assets through the bank’s app. During the 2 redesign I worked on both the direct Bitpanda product and the bank ready variant used inside Raiffeisen. The same system that later contributed to 53% user growth (3.4M → 5.2M) and more than €140M+ annual digital revenue across the period.
AI-Based Prediction Market
Benker
Benker is a digital banking platform built on blockchain and tailored for secure, multi currency account management across Europe. As design lead and hands on designer, I redefined onboarding and transaction flows, which increased digital transaction volume by 150%, improved KYC success rates by 30%, and reduced time to first successful transaction by 50%.
AI-Based Prediction Market
OnRobot
Built a tablet-first HMI in 12 weeks to eliminate teach-pendant scripting and spreadsheet math. In pilots it cut first-time setup by 42%, reduced input errors by 63%, slashed training time by 75%, and drove a +36 NPS among operators.
AI-Based Prediction Market
SportsGambit
SportsGambit is a decentralized prediction market leveraging advanced AI to give you an edge. The platform's core feature is the ability where a user can build, train, and deploy custom prediction agents. These autonomous agents continuously learn and improve, analyzing specific sports or events to deliver a curated feed of high-probability outcomes directly to your dashboard. This eliminates manual research and allows for swift, one-tap wagering, combining AI-driven insights with the security and transparency of the blockchain.
Challenges
The primary challenge was foundational: how to build user trust in a novel, autonomous technology. Prediction markets are already complex. Adding layers of "black box" AI agents and blockchain security created a significantly high barrier to entry for the average user.
The key hurdles were:
Simplify AI Training: The process of "training an AI" had to be demystified for non-technical users who simply wanted an edge in their wagers.
Build Trust: User skepticism had to be overcome by making the AI's performance transparent and its actions understandable.
Design for Versatility: A single, intuitive interface was needed to cohesively handle diverse and fast-moving markets, from sports and crypto to global events.
Enable Speed: The design had to allow users to act instantly on high-confidence, time-sensitive picks, removing the friction of manual research.
Key UX Methodologies
To tackle these specific challenges, a focused set of UX methodologies was chosen:
Jobs-to-be-Done (JTBD): This was the guiding framework. It was clear users weren't "hiring" the product to configure an AI; they were hiring it to win more bets with less effort. This "job" informed every design decision, shifting the focus from complex features to simple, valuable outcomes.
Progressive Disclosure: To solve the high barrier to entry, this method was critical. Turned “training an AI” into a short, step by step flow in plain language with helpful presets. Users always know where they are and why, which lowers drop-off at setup.
Lean UX & Iterative Sprints: The project was run from concept to MVP in a series of fast, iterative sprints. Given the novelty of the product, this approach was essential. It allowed the team to get a functional version in front of users quickly, test core assumptions, and adapt based on real-world behavior rather than internal speculation.
Mental Model Alignment: The design process focused on using plain language and smart defaults. By avoiding technical jargon ("Epochs," "Learning Rate") and instead using simple explanations, the system was designed to match the user's mental model of "making a good pick," not a data scientist's model of "building a neural network."
The Process
The design was led from concept to MVP, focusing on core user flows and building a system that was both scalable and easy to use.
Homepage as a Decision Surface: The dashboard was redesigned to function as a "decision surface," not a generic landing page. It was built to immediately answer the user's core question: “What should I do today?” This was achieved by surfacing AI-generated picks with clear confidence signals and one click bet actions.
Guided Agent Creation: The agent setup was rebuilt as a clear, step-based flow. It uses smart defaults, plain-language explanations, and light progress cues so users always understand where they are in the process and why each step matters.
Unified Market View: A single, unified "market card" pattern was introduced. This card contains all critical info live odds, a clear Yes/No action, and return previews allowing users to switch between categories like sports and crypto without having to relearn the interface.
Research & Problem Solving
The design strategy was centered on leveraging the rise of AI agents and DeFi, with a relentless focus on simplicity, personalization, trust, and speed.
Problem: How do you make complex AI training accessible?
Solution: A modular workflow was designed with smart defaults and guided training options. This approach successfully turned a potentially complex configuration task into a few simple choices.
Problem: How do you get users to trust an AI's predictions?
Solution: Transparency was designed into the system from the start. Transparent performance leaderboards and clear confidence scores for every pick allow users to see and validate an agent's track record before acting on its advice.
Problem: How do you help users find value in a noisy, fast-moving market?
Solution: The personalized dashboard was designed to auto-populate with a curated list of high-confidence picks. This filters out the noise, allowing users to focus on effortless, one-tap actions.
The Outcome
The user-centric design approach was validated by strong post-launch metrics:
High Quality Activation: 35% of new users successfully completed the entire core activation loop (created an agent, received a pick, and placed a wager) within their first 24 hours. This confirmed the guided setup's effectiveness.
Cutting Through the Noise: The personalized dashboard became the platform's primary engagement point. Data showed that 70% of all wagers placed by active users originated from its curated, 'one-tap' AI recommendations, validating the design's focus on moving away from manual research.
Behind the Decisions: Reflections & Trade offs
On SportsGambit, everything came down to the few seconds before someone decides to put money on the line.
We knew two things: more “AI explanation” doesn’t automatically mean more trust, and less information doesn’t automatically mean speed. The job was to design a moment where someone glances at a card and instinctively knows: what the bet is, how confident the system is, and whether they’re comfortable acting on it now.
That’s why we went decision-first: one clear action, then optional depth. The team killed a lot of visually interesting concepts because they failed the simple test: “Can a new user land here, understand the story in under five seconds, and still feel like they’re in control of the choice?” If the answer was no, it didn’t ship.
