All projects
Designed Genius Answering, a complex AI assistant that operates as a 24/7 receptionist and dispatcher — managing calls and messages, responding instantly to customer inquiries, and turning interactions into booked jobs directly within the user’s calendar.
Problem
Genius Answering is a virtual assistant that communicates with customers through calls and messages. The most important user interactions aren't happening inside the product UI — they are happening in conversations between customers and the assistant.
01
Designing the system behind the conversation
This changed the design challenge from simply creating screens for users to designing the assistant experience and the control layer around it — helping businesses configure, understand, monitor, and trust an AI that acts on their behalf.
02
The interaction happens outside the platform.
The control happens inside it
However, while customers interact with Genius through calls and messages, business owners still need to manage and monitor the assistant’s performance within the platform. The old Genius Answering interface wasn’t effectively communicating the assistant’s value, capabilities, and statuses.
Goals
🎯
User goals
Get started with minimal friction and set up fast
View plan details, plan usage, and available features at a glance
Easily understand and configure features based on their needs
📊
Business goals
Improve activation experience, reduce drop-offs and increase retention
Reduce support requests and encourage self-serve upgrades
Increase feature adoption
Discover & define
01
Understanding the assistant’s behavior
I started the research by auditing the existing experience and assistant behavior to understand where the product was working and where gaps existed. I reviewed sales calls recordings in Gong with business owners and admins to understand their questions and expectations around Genius Answering capabilities.
Customer feedback on assistant–customer interactions
🎧
End-customer conversations
01
Business-specific answers
Many conversations required custom responses defined by the business. Without them, the assistant could struggle to understand the context and provide an accurate response.
02
Redirect to a human
Customers recognized the assistant as AI and asked for a real person — though sometimes Genius managed to keep the conversation going :)
03
Not all requests fit the logic set up
For example, a new estimate could represent a lead, while scheduling a new job could represent a new job, requiring additional conditions.
Free trial starts dropped by approximately 60%
02
Understanding the management experience
I then focused on how this assistant behavior translates into the platform experience businesses need to manage:
↳ Analyzed FullStory sessions recordings, Amplitude analytics and support tickets to identify drop-offs, key flows gaps and engagement patterns.
↳ Mapped how the assistant’s capabilities and limitations should be represented in the platform to support business management workflows.
💡
Key insights
01
Activation funnel gap
A significant decline in free trial starts indicated a breakdown in the activation funnel. Users were not motivated enough to start the trial.
02
Feature configuration
Increasing complexity and a growing number of capabilities led to setup abandonment and limited feature engagement.
03
Dashboard transparency
Users lacked a clear understanding of their current plan, feature statuses, and the overall state of their Genius Answering.
Research conclusion
For business owners and admins, the management experience determines how easily they can configure the assistant, understand what it can and cannot do based on their plan and track how it handles real-world requests.
The way the assistant’s capabilities and limitations are set up and managed within the platform directly shapes the success of end-client interactions — ultimately creating value for the business through the AI assistant.
Design & deliverables
Onboarding
01
I introduced a clear, concise value proposition upfront and designed a simple 4-step guided onboarding flow that focuses only on what’s essential to get started with Genius Answering. To reduce effort and friction, most inputs are prefilled using existing account data, so users are mainly confirming and moving forward — making setup feel faster, more intuitive and less manual.
A key addition was a dedicated Genius AI test modal, where users can interact with the agent in real time. This helps them experience the product before committing and builds trust early in the journey.
Dashboard
02
I standardised the layout so it feels familiar and easy to scan, with a clear header and consistent placement of core elements across the page.
Trial period
The trial label now clearly indicates the number of days remaining, with a contextual upgrade prompt. The messaging stays visible without interrupting access to core functionality.
Feature control overlay
The feature control overlay highlights the key capabilities of the feature. This overlay is essential, as activating both toggles — calls and
messages — directly increases usage and revenue.
System feedback
Configuration
03
The redesigned configuration experience makes the assistant’s capabilities easier for businesses to understand and manage. Capabilities are grouped into clear, logical sections, helping users quickly see what the assistant can do and configure it according to their needs.
The experience also makes limitations more transparent. Upgrade entry points are placed in context, allowing businesses to see which capabilities are available on their current plan and preview what additional functionality could unlock.
Results & impact
Measured over 4 months post-launch across 1,400+ active accounts, the redesign drove meaningful improvements in discoverability, adoption and engagement.
+48%
Growth in core engagement
Total increase in monthly conversations
37.9%
Trial conversion rate
Across 4 months of post release
2x
Adoption of Genius messages
Genius messages enabled usage doubled
See more
Creating a unified design language for a growing product
SaaS
Web App
In progress
Workiz Phone — UX Improvements
SaaS
Web App
Shipped