Duos AI
Overview
Duos AI is an AI-powered product focused on helping users prepare for interviews through practice and feedback.
I am working across monetisation, experimentation, visual and product flow to improve conversion and enable faster iteration..
Problem
The product had strong core functionality, but:
Monetisation flows were not optimised for conversion
No structured experimentation system (A/B testing)
Paywalls were difficult to iterate on quickly
Limited visibility into user behaviour and feedback across flows
Hypotesis:
If we improve how and when users encounter monetisation, and enable faster experimentation, we can increase conversion and learn what actually drives it.
Solutions
Focused on two areas:
1. Monetisation clarity & timing
Make upgrade moments more contextual and aligned with user intent
2. Communication with users from different area of flow in one CRM System
Discovery & direction
Building on existing research, I focused on identifying which features users valued most and where monetisation could be introduced without disrupting the experience.
I analysed user behaviour patterns and key triggers to understand when users were most engaged and where monetisation would feel natural and useful.
Early on, I also proposed evolving the monetisation strategy to strengthen the product’s value proposition - particularly in the context of future investor conversations.
Contribution
I presented several monetisation directions to the team, focused on aligning upgrade prompts with high-intent moments in the user journey based on research like competitors, other screens, articles about mental models, my knowledge and common sense.
After alignment, I took ownership of designing and implementing paywall experiences.
The goal was to ensure monetisation felt like a natural extension of value, rather than an interruption.
Draft
Before I was even thinking about UI, I created a draft paywall with AI to quickly discuss with the team that we are happy with the direction, information, and proposal.
Versions
Created a new version and presented it to the team for discussion.
Each team member contributed their perspective, which I filtered and incorporated into the updates.
Testing
A/B testing setup
(On one paywall to reduce timing of implementation)
Introduced structured experimentation approach
Designed multiple paywall variations for testing
Enabled faster iteration cycles based on results
Implementation
Paywall implementation (Adapty)
Translated designs from Figma into Adapty manually
Built flexible paywall configurations for testing
Reduced dependency on engineering for updates
Adding legal part tot me approves - Terms, Privacy Policy, Return
Feedback
Discovery & direction
When I joined, we didn't have any communication with users for feedback, so I began working on this aspect as well. I am still organising other parts of the flow, such as the report bag in the profile or during the interview process, but the second area I focused on after my first interview was the mood tracker. Here is the old version. Essentially, you could only show how you felt, and that was it.
Impact
Now, our feedback approach looks like this, and we have three options for communication:
1) If a user is frustrated after an interview, we acknowledge their feelings and sympathise, while also gathering feedback to understand how we can improve the process in the future.
2) If the feedback is neutral, we simply acknowledge it.
3) If the user is happy, we take the opportunity to gather an App Store rating and ask for a review, which is available in your app after the user logs in for the first time.
Additionally, I have added animation and fun elements to encourage more emotional interaction.
Other things
I work on the design system by maintaining and publishing components, ensuring everything is responsive and consistent throughout the app.
