Omny provides an AI platform that aims to expand Amazon profits and simplify operations for consumer brands and their agencies. The platform offers a suite of tools that cover various aspects of Amazon selling, including advertising, P&L and analytics, content optimisation, and supply management.

‍

The platform seeks to help brands increase revenue, lower costs, streamline operations and gain more control over their business. Omny's features include automated advertising optimisation, real-time analytics, AI-driven content enhancement, and inventory management, all within a single platform. The company also offers flexible services, including managed solutions.

The challenge

Omny’s core challenge was creating the right level of user engagement with AI suggestions. The goal was to find a balance where users would carefully consider AI suggestions without the process becoming too cumbersome or inefficient.

‍

In the initial design for product content management, we prioritized efficiency by pre-populating AI suggestions for product benefits and use cases, requiring minimal user clicks.

‍

This approach, however, led to users skimming past important decisions, and even though the AI had a 90% accuracy rate, occasional errors damaged user trust. Users felt betrayed by incorrect suggestions they had not properly reviewed.

‍

Users were holding Omny accountable for suggestions they hadn't properly reviewed, creating a negative experience and reducing trust in the AI.

‍

The original design created a tension between convenience (instant AI suggestions) and accuracy (careful user review).

Our final solution

We reframed the AI-human relationship to encourage collaboration instead of presenting AI outputs as decisions.

‍

The design was changed so that the area for the final product benefits and use cases was kept empty, with AI suggestions on the side. Users then had the responsibility to click on the suggestions they felt were correct, populating them in the list.

‍

This new design created a heightened awareness about how users were interacting with AI suggestions.

‍

The solution aimed to balance efficiency with user engagement by:

• Providing a blank canvas for the final content, separating suggestions from decisions and giving users control over the final output.
• Presenting AI suggestions in a distinct side panel with a simple click-to-add mechanism and clear visibility of the selection process.
• Maintaining efficiency through easy selection while adding moments for review.

Key design decisions

Our design process began at the information architecture level, even before we analyzed the user interface. We analyzed raw data patterns and identified meaningful relationships between different data points, thereby presenting a processed version to the user.

‍

The design includes consistent visual patterns for AI-driven suggestions and insights through the use of two primary colours: purple to indicate AI-generated content, suggestions, or automated actions and blue to represent user inputs, manual overrides, or direct user control. This color system is consistently applied across various interaction patterns, such as in the budget allocation interface and budget strategy timeline.

‍

In the budget allocation interface, for example, purple AI icons display AI-suggested weekly budgets, and the cells transform to blue when users override these suggestions. This color shift provides visual feedback of the transition from AI to user control.

‍

Similarly, in the budget strategy timeline, a blue line represents the user-controlled available budget, while a purple line shows AI-managed spending.

‍

For product content management, purple tags display AI-extracted benefits from user reviews, while blue tags show manually added product benefits. The system also indicates the scope of AI analysis.

‍

Another important focus was on creating data micro-journeys. Each screen element strategically anticipates and answers the questions that arise as users process the preceding information. For example, the design of the weekly advertising budget strategy management interface included how the user would decide the amount of budget to allocate and what KPIs could be shown to help with this.

Impact and key learnings

• The new design increased user awareness about how they were using AI suggestions.
• Users became more forgiving of occasional AI inaccuracies because they were active participants in the selection process.
• The design strengthened trust in AI capabilities while improving AI-human collaboration.
• The key learning was that the success of AI interfaces depends not just on the quality of AI suggestions but on how users interact with them.
• It is important to create a journey with the right amount of friction to ensure users engage with AI suggestions without the process becoming too cumbersome.
• Maximum efficiency (instant AI suggestions) conflicted with maximum accuracy (careful user review), and it is important to find a balance between these two goals.