Building an AI agent beta product from scratch.

Gaddr is a startup and comprehensive AI agent platform built to automate business processes across project management, marketing, HR, and more. I clarified user needs and helped design for the beta product launch.

Client

Client

Gaddr

Industry

Industry

AI Agents B2B SaaS

Company size

Company size

50+

Team size

Team size

5 designers

Problem

To create a beta product amid unclear vision, shifting requirements and lack of structure, causing misalignment and inefficient feature prioritization.

I introduced design processes and key UX artifacts to align the team, clarify the product vision, and shape early ideas into a structured, testable beta product.

Results

80% clarity rate. Beta users understood the app’s purpose and core features, indicating that the initial UI effectively communicated the product’s value.

60% of beta users tested at least one AI automation feature, showing strong early engagement with the platform’s key functionality.

25% of users provided actionable feedback for future improvements, demonstrating an active and invested user base helping to shape the product roadmap.

Average session length: 8–12 minutes per user during beta testing, reflecting meaningful interaction time, suggesting users found value in exploring the app.

85% of users rated the prototype as “usable” in usability tests, confirming that the design met a high standard of usability even at this early stage.

80%

Understood apps purpose

25%

Provided actionable feedback

8-12 min

Average session length

Process

Research & Analysis: I benchmarked AI agent platforms and interviewed internal stakeholders to understand user needs, technical constraints, and UX patterns. Key insights, like the need to support both chat and workflow-based input, shaped our persona, design direction, and testing priorities. A user persona was created to represent a key user.


Establishing a Design Process: I defined the MVP, created the information architecture and helped develop a sitemap to organise the product content and map key user flows to create a clear navigational structure. I introduced regular design reviews and stakeholder meetings to align the team, surface issues early, reduce friction in collaboration and build a shared understanding of the product structure and user journey.


Wireframing: I created low-fidelity wireframes to define layout, user flows, and content hierarchy. Key areas included:

• A customizable dashboard to surface key task information.
• Task orchestration and agent chat for managing automated workflows.
• Login, signup, and settings to support smooth user access and control.


UI & Design System: I created a design system and component library to bring consistency and scalability to the UI. Feedback refined interactions, improved usability, and shaped a navigation model that scaled across the product. The idea was to balance clarity and customization, enabling users to quickly access key information and tailor their view to fit their needs.

Main dashboard with customizable widgets showing upcoming events, tasks etc. Priority tasks and calendar highlighted giving user most relevant info at a glance.

Chat pane opens on right side, giving quick access to chat input. Widgets resize to reduced screen size keeping priority info visible for easy multitasking.

First Iteration:

The agent chat pane as a pop up window overlay over the main dashboard to focus user attention, as shown below. Based on feedback, we shifted to a less intrusive layout allowing users to view the dashboard and chat simultaneously, better supporting multitasking.

Conclusion

I handed over a solid foundation of assests and processes, including a detailed UI, wireframes, a developing design system, and clear documentation of user flows and requirements. These assets were designed to ensure consistency, speed up decision-making, and support future design work and iteration.


The next steps would be to continue testing the designs with users to validate key flows and identify early pain points and analyse metrics to refine interactions based on feedback before full scale development.


My goal was to leave the team with a clear, scalable foundation to build on. I was proud to see the project brought to life from an unclear concept to a tangible product and UI setting the stage for future evolution.