AI UX Design Principles at Work: How We Helped Accern Build Rhea

According to McKinsey, 75% of companies are using AI in at least one business area — and that number keeps growing. But there’s a catch: most AI-powered platforms still feel confusing, cold, and hard to trust. Why? Because great AI UX design is rare. For AI tools to actually work, they need interfaces people understand, built by UX designers who deeply consider user behavior, intent, and feedback.

That’s where Lazarev.Agency stepped in to help Accern rethink how users interact with their financial research web app, Rhea. We designed its UX foundation and visual design, making the AI ​​interface accessible and useful for different types of analysts — here's how.

Key Takeaways

• Rhea is about UX that feels natural, not just novel
• We implemented a UX pilot guided by user behavior, not assumptions
• The solution introduces adaptive inputs for all user types
• The tool enables real-time workflows, replacing tab-jumping
• Transparency is a cornerstone of our work with Rhea

More Than Ordinary Chatbot: Rhea’s Adaptive AI UX Design

When Accern set out to build a research tool powered by AI, they didn’t want another chatbot with charts. They needed an AI assistant that actually helps financial analysts do their work effortlessly — faster, deeper, and without friction.

The heart of Rhea is a prompt-based AI model. But we knew early on that prompts alone weren’t enough. So we built a hybrid interface: part conversational AI, part modular GUI — empowering UX researchers and product teams to co-create an adaptive experience. 

The result? A system that adapts to how different users think and act — whether they prefer typing out commands, clicking buttons, or uploading text data.

• Widgets respond to user intent. 
• Tooltips adjust dynamically. 
• Even the input field transforms based on context — at one moment it’s a chat box, the next it’s a command center.

"Our goal was to make the AI feel like a reliable research partner, not a machine that dumps data."

{{Kyrylo Lazariev}}

Split-Screen Research & Reporting

One of the biggest wins from this AI UX design project was a key behavioral insight: researchers often copy-paste between tools. We eliminated that.

With Rhea’s split-screen mode, users explore data on the left and build their reports on the right—in real time, in one place.

"It’s not just faster. It’s cognitively lighter. Instead of juggling tabs, you stay focused."

{{Kyrylo Lazariev}}

Text, charts, quotes, even references — everything can be inserted, edited, and arranged without breaking the flow.

Smarter Prompts, Smoother Conversations

Designing natural language interaction that actually delivers, goes beyond basic input parsing. Rhea’s AI UX design is about helping users shape their thoughts in a way the system can work with.

When users aren’t sure how to phrase a question, Rhea helps refine it. Auto-suggestions, follow-ups, clarification prompts — it’s like an AI that knows how to ask better questions with you, not just for you.

This keeps users moving, especially when they hit a dead end. And it builds confidence: even non-technical analysts feel fluent inside Rhea.

AI UX Design Principles: Visibility, Traceability, Trust

Great AI UX gives users clarity over how AI works, keeps them in control, and supports collaboration rather than automation. It adapts to context, handles errors transparently, and reveals complexity only when needed.

For financial analysts, answers alone are barely enough. So we built transparency right into the interface. Every AI-generated result comes with visible references — think Crunchbase, PitchBook, news links. Each insight is traceable, clickable, and ready to be pinned to a report.

"Transparency is the difference between rejection and trust. In Rhea, we made every AI output traceable — so users can verify, act, and move forward with confidence."

{{Kyrylo Lazariev}}

From Answers to Actions

Rhea doesn’t stop at giving answers. Instead of dropping plain text, it responds with live, interactive widgets:

• Graphs with drill-downs
• Tables you can edit
• Charts you can export
• Blocks you can add to reports

Each element of Rhea’s AI UX design is actionable, made to turn passive results into active next steps. That’s how we turned a prompt field into a productivity engine.

Multipurpose Input = Maximum Control

We reimagined the chat input as a multi-modal command line. It can:

• Accept queries
• Trigger workflows
• Upload datasets
• Schedule emails or alerts

It’s a quiet powerhouse. No dropdowns. No clutter. Just one field that adapts to what the user needs.

AI UX Design That Disappears Into Work

In complex tools, good design feels invisible. That’s what we aimed for with Rhea:

• Context switching is gone. You stay in one environment.
• Feedback is instant. Microinteractions guide attention.
• Everything is modular, editable, and contextual.

"The more seamless the UX, the more powerful the AI feels."  

{{Oleksandr Koshytskyi}}

Rhea’s interface is the difference between a smart tool and a tool you want to use every day.

Form, Focus, Flow: Rhea’s Visual Logic

Rhea was designed for the hands of professionals making high-stakes decisions.

The UI leans into a dark theme, not for aesthetics, but for focus. Sharp contrast and restrained color usage draw the eye where it needs to go. Nothing shouts. Everything guides. It’s the kind of digital design that says: this tool means business.

Typography is tuned for clarity. Open shapes, balanced weight, and neutral tones make data — whether text or numbers — feel approachable and effortless to read. When users spend hours in a system, readability stops being optional.

"Hardly anyone can explain good typography. Yet every user feels the difference it makes."

{{Oleksandr Koshytskyi}}

The interface follows a modular structure. Panels expand when needed, collapse when not. The split-screen layout organizes space in a way that reduces cognitive drag. That’s a UI decision with UX-level impact.

Details matter: 

• Microinteractions respond without delay.
• Translucent layers add depth without distraction.
• Rounded edges soften an otherwise high-precision environment.

Behind the Scenes of Our Design Process: What Made It Click

We didn’t start our AI UX design with screens. We started with real users. Through interviews, workflow mapping, and testing prototypes with real analysts, we learned what slows them down — and what helps them move.

Here’s how our design process unfolded:

• Step1: User Research & Discovery
• Step 2: Pain Point Mapping
• Step 3: Workflow Mapping
• Step 4: Prototype & Test Early
• Step 5: Iterate Based on User Feedback
• Step 6: Design for Adaptability
• Step 7: Build Around Real Use Cases

That’s how Lazarev.Agency came up with lenses: pre-configured data sets analysts could switch between. That’s how we built inline CTAs, enabled dataset linking, and added email-based updates.

Everything in Rhea was designed to meet people where they are and create pathways to take them further.

AI Tools We Use for UX Design at Lazarev.Agency

At Lazarev.Agency, we leverage the latest AI tools to optimize our UX design process, ensuring seamless and intuitive experiences. Here’s a look at the key tools we use to generate wireframes:

1. Figma & Sketch – UI design & prototyping
2. These industry-standard design tools are enhanced with AI-driven features to automate tasks, suggest design improvements, and allow us to quickly prototype and iterate on user interfaces.
3. Maze & Useberry – User testing & feedback
4. We use AI-powered platforms like Maze and Useberry to conduct interactive user tests and gather actionable feedback in real-time. These tools help us refine the UX based on actual user behavior and interactions, making sure our designs meet real needs.
5. Hotjar & Crazy Egg – Behavioral analytics
6. These AI-driven analytics platforms track user behavior, identify friction points, and provide heatmaps and session recordings. With this data, we can pinpoint areas of improvement and enhance the user journey for maximum engagement.
7. Notion & FigJam – Collaborative design & workflow mapping
8. For ideation, brainstorming, and workflow mapping, we use Notion and FigJam. These tools help our team collaborate efficiently, and AI features allow us to automatically organize, suggest relevant resources, and optimize our design processes.
9. ChatGPT & Claude – AI-enhanced interaction design
10. To create intelligent, responsive conversational interfaces, we integrate tools like ChatGPT and Claude. These models help us simulate and refine natural language interactions, ensuring our AI-powered designs can seamlessly guide users through complex tasks.

We believe great design happens when strong UX thinking meets the right tools — and we’re always exploring what’s next. While we rely on trusted platforms like Figma, we also experiment with new AI design tools to go beyond the traditional toolset.

Here are some of the AI tools we actively explore and learn from:

• Galileo AI – A powerful AI UI generator that turns text prompts into editable high-fidelity mockups. Great for early-stage ideation and quick visual output. Explore Galileo
• V0 – Generates production-ready UI code using Tailwind and shadcn components, helping bridge the gap between design and development.
• Uizard – Converts hand-drawn sketches and text into wireframes and editable UI mockups. Perfect for fast concept validation and design sprints.
• Magician (Figma plugin) – Adds AI-powered microcopy and visual elements right into Figma, speeding up creative decisions.
• Diagram’s Automator – Streamlines repetitive tasks in Figma, freeing up time for more meaningful design work.
• Relume – Generates sitemap structures and wireframes from simple prompts, helping kickstart early UX concepts.

Many tools offer a free plan, standard plan, or paid plan. Depending on the team’s needs, a free account might be enough for early-stage design, while a pro plan or paid subscription unlocks the full range of features:

‍

‍How to Turn Design Ideas Into User-Friendly Experiences for Your Business

Artificial intelligence is complex, but using it shouldn’t be. The UX of Rhea doesn’t aim to impress—it’s designed to disappear, becoming so intuitive and fluid that you forget it’s even there.

With Rhea, we transformed a powerful AI backend into a tool that actively supports users in their decision-making. This is the essence of great AI UX design: making the complex simple and helpful.

Are you ready to create AI tools your users will actually enjoy? Or redesign an existing product to align with real user behavior? Let’s talk and come up with a design vision for your company.