AI-powered tool for Tax Auditors
User Interviews, Prototype-First Approach, System States & Edge Cases, Workflow Mapping, Dev-Handoff
Overview
During tax season, CPA firms are overwhelmed by the manual grind of managing client data collecting forms, keying numbers into systems, and reconciling information across spreadsheets. What should be simple review work often turns into hours of repetitive data entry and verification.
SignalsHQ was designed to relieve that pressure. The platform introduces a unified, trustworthy workspace where reviewers can securely manage client documents, verify data, and access clear, approved summaries. By automating the tedious parts of data handling and surfacing only what matters, it helps CPAs focus on judgment, not paperwork.
My Role
I was responsible for shaping the product experience from a written PRD into a functional, testable design system. I translated early product requirements into interactive prototypes, defined the core workflows. Working closely with the product and engineering leads, I iterated on structure, usability, and trust cues to ensure the design balanced automation with human oversight.
Deliverables
Document Collection & Management
A unified, compliant workspace for adding, finding, and triaging client-year documents with defensible versioning, and accurate reclassification. Enabled scalable intake through client portal intakes.
Extraction Suite
Created the end-to-end experience for automated document classification, structured data extraction, and human-in-the-loop review — enabling confidence tracking and efficient correction flows.
Neural AI (Client Intelligence Layer)
Developed the Client Intelligence layer (1040 MVP) to generate trusted rollups, provenance-backed drill-downs, and three-year category trends all derived from approved extracted data giving CPAs confidence in the numbers they see without rebuilding spreadsheets or chasing forms.
User journey flows, high-fidelity screens, rapid prototypes validated through usability checks. Supported by workflow diagrams, and component-level documentation for seamless handoff.
And mostly turning what to what if?
Thanks for Visiting!