Balto

Role: Product designer
The challenge: Determine whether Balto's proprietary AI engine could support a scalable coaching and QA product, then translate the opportunity into a cohesive enterprise platform.
Top outcome: Established the foundational product architecture for Balto's Real-Time Coaching platform through rapid exploration and high-fidelity prototyping.

From desktop widget to enterprise platform

Balto had a powerful real-time AI engine, but its value was largely confined to a narrow desktop widget. The company was exploring whether that technology could become the foundation for a broader coaching and QA platform. I joined as the first designer on the initiative, working with product and engineering to explore the opportunity, test concepts with customers, and translate the strongest ideas into a cohesive product experience.

Low-fidelity wireframes mapping out real-time call center coaching workflows — Exploring the AI engine's capabilities through low-fidelity structural flows and wireframes.

Architecting the coaching ecosystem

Expanding into call center operations required designing for distinct user roles simultaneously: agents handling live calls, supervisors monitoring teams in real time, and administrators managing QA rules.

Working alongside the product manager, I mapped complex feature requirements into discrete interaction patterns. We established core workflows for real-time agent prompts, automated event triggers, and supervisor intervention tools built to handle live, high-stress environments.

Add new alert modal interface for setting up rule-based real-time AI triggers — Core interface patterns defining agent assistance, real-time supervisor alerts, and administrative rule creation.

Call monitoring card showing flagged compliance alerts, action controls, and a live call transcript snippet — Core interface patterns defining agent assistance, real-time supervisor alerts, and administrative rule creation.

High-stakes, real-time feedback

Call center operations move fast, and cognitive overload is a constant risk. When the AI detects a critical moment on a call, the supervisor needs immediate context to decide whether to listen in, send a direct coaching tip, or take over the call entirely.

I created high-fidelity prototype flows to stress-test these high-stakes interactions. We refined density, visual hierarchy, and status indicators so supervisors could evaluate agent performance and trigger interventions in seconds without disrupting ongoing operations.

Balto web application interface showing configured alerts table alongside a live supervisor feed and chat sidebar — High-fidelity prototype flows evaluating live supervisor interventions and post-call performance scoring views.

QA Scorecard builder interface displaying call evaluation categories, percentage weights, and scorecard configuration settings — High-fidelity prototype flows evaluating live supervisor interventions and post-call performance scoring views.

Aligning stakeholders and scaling the vision

To keep engineering and executive leadership aligned, I facilitated rapid, daily review cycles using detailed Figma prototypes. Working closely with the lead designer, I helped integrate evolving visual patterns and brand standards into a unified component library.

The resulting product architecture transformed Balto from a point-solution widget into a complete real-time coaching platform, providing the foundational product strategy needed to drive their enterprise expansion.