Redefining human transactions with AI & blockchain
UX/UI Design, Project Management, Product Research
Thomas Dyrks
Following a successful ICO launch, VAIOT set out to evolve its mobile app beyond the initial concept. The goal of our partnership was to translate their ambitious vision: combining the capabilities of a lawyer, a blockchain developer, and an AI personal assistant, into a cohesive, user-centric product experience. Building on insights from the first iteration, our challenge was to design and prototype a product that could validate product–market fit and support VAIOT in securing their next round of funding.
My role
As a Product Designer, I was responsible for:
Dynamic technology landscape
The AI voice chat technology was being developed simultaneously with the app, meaning final capabilities were unknown during the design phase. It required close collaboration between design and engineering teams to ensure flexibility and readiness for multiple rapid iterations.
Designing for the invisible
Unlike visual interfaces, voice interactions leave little space for visual cues or affordances. We had to anticipate edge cases, misunderstandings, and user hesitations, crafting fallback flows and natural conversation patterns that felt helpful, not robotic.
Context management and state awareness
In a voice-led interface, users expect the system to “remember” context, even across multiple turns or intents. Without robust state management, the experience quickly feels disjointed. Our challenge was to design a dialogue model that handled handoffs, corrections, and multi-step tasks without requiring users to repeat themselves.
Latency and response handling
AI-powered voice experiences are only as good as their response speed. Delays of even a second can break the illusion of fluid conversation. Designing around latency, both anticipated and unexpected, meant carefully orchestrating UI feedback, fallback voice prompts, and graceful error handling that preserved user confidence.
The project kicked off with an in-depth exploration of VAIOT’s renewed vision and strategic goals. With initial funding secured, the focus shifted toward developing a version of the product capable of validating real user sentiment. To align on direction, we facilitated a series of workshops to deconstruct and define the core elements of the product’s value proposition. In the absence of direct competitors, we turned to adjacent spaces - chat-based interfaces and crypto payment apps for inspiration, balancing proven patterns with space for meaningful innovation.
One of the most significant milestones was designing the chat interaction model. Unlike traditional interfaces, voice-based interactions introduced a distinct set of challenges. To ground our approach, we explored similar solutions through technical specs, academic research, and industry reports. Understanding the technological feasibility of a voice assistant was as critical as defining its user experience. We dedicated substantial time to exploring emerging voice UI technologies, ensuring our design choices were not only ambitious but also implementable. This phase concluded with a deep tech audit that shaped the constraints and opportunities of the product’s conversational layer.
To better understand how users would interact with a voice-based assistant, we conducted early-stage user research using custom-coded dummy conversations. These scripted flows simulated key interactions without requiring a fully functional backend, allowing us to test assumptions quickly and gather qualitative insights. Our first use case focused on purchasing car insurance.
That was a complex, multi-step process that provided a rich scenario for testing conversational dynamics. By observing participants as they navigated the simulated dialogue, we uncovered critical expectations around tone, pacing, and information clarity. We also identified pivotal moments in the flow, such as:
where trust, transparency, and guidance became essential. These insights helped us refine the conversational architecture, highlighting where users needed reassurance, autonomy, or simplification. The research not only validated core assumptions but also shaped the foundation for scalable, user-centric dialogue design in future use cases.
With core user expectations mapped out, we shifted focus to elevating the interaction from functional to fluid. Using insights from our simulated tests, we iterated on tone, flow, and phrasing - transforming rigid system prompts into more natural, human-like exchanges. We paid particular attention to micro-interactions: pauses, confirmations, and fallback responses that would either build or erode user trust. The goal was to create a voice assistant that felt not only competent, but approachable and intuitive, even in complex scenarios like insurance selection or crypto transactions.
Once the core conversational experience was in place, we broadened our focus to cover the surrounding systems that would complete the product. While not central to the initial pitch, areas like payment flows, data management, and onboarding were essential to deliver a seamless and trustworthy user experience. Particular attention was given to how users would authenticate, provide sensitive information, and review key decisions.
To support both users and the marketing team, we designed a clear and engaging onboarding experience that introduced VAIOT’s value proposition in simple, accessible terms. Given the complexity of combining AI, voice interaction, and blockchain technology, we developed an explainer that broke down the product’s capabilities and benefits into intuitive steps.
This not only helped reduce friction for first-time users but also served as a strategic asset for marketing, ensuring consistent messaging across product and promotional channels. The result was a smoother onboarding flow and greater confidence from users during their first interactions with the assistant.
Start with the core
Before diving into edge cases or peripheral features, it’s critical to nail the core experience. In our case, it meant getting the main conversation flow right - making sure users understood, trusted, and enjoyed the interaction.
The smallest details matter
In voice and AI-driven interfaces, subtle choices (wording, tone, timing, even a pause) can dramatically influence how users perceive and trust the system. Attention to micro-interactions often makes the difference between a tool that feels robotic and one that feels intuitive.
Design is iteration, not perfection
Working on emerging technologies means accepting change as a constant. Rather than aiming for a perfect first version, we embraced a culture of continuous iteration: prototyping fast, testing early, and refining based on feedback.