Defining Product Direction: Smarti was positioned as a cloud-based OCR solution focused on extracting structured and semi-structured data from business documents. The intent was to provide enterprises with a centralized system that could support document ingestion, review, and correction workflows across teams, while prioritizing clarity and user control over aggressive automation.

Information Architecture Decisions: Research and workflow observation showed that agents frequently switched between data views to verify extracted values, increasing cognitive load and slowing reviews. The information architecture was designed to reduce this context switching by clearly separating review stages while keeping critical data accessible at all times. The five-tab structure established clear functional boundaries between stages of document review, allowing users to focus on one task at a time without losing orientation and making it easier to return to specific sections during exception handling.

At a macro level, each section focused on a distinct category of information required during document review. This structure allowed users to move through verification tasks systematically rather than scanning dense, all-in-one views, with priority given to the most frequently validated fields.

At a micro level, content within each section was structured to support fast scanning and error detection. Consistent grouping, spacing, and alignment reduced visual noise and helped users quickly compare extracted values with source documents, which was critical for detecting mismatches in high-volume workflows.

Interaction and UI Design: UI decisions were guided by usability, accessibility, and readability. Tabs were placed at the top of the interface to align with common scanning behavior and provide predictable navigation. Extracted data was kept editable at all times, allowing users to correct inaccuracies without disrupting their workflow. Rather than locking fields to speed up completion, the interface prioritized transparency and control to support compliance-driven decision-making.

Visual Design Choices: The visual design extended existing brand colors while adapting them for improved legibility and accessibility. Additional colors were introduced selectively to represent system states and document statuses, helping users quickly understand context without overwhelming the interface. Visual consistency was maintained to ensure the product felt professional and reliable.

Prototyping and Collaboration: Design concepts were translated into interactive prototypes through close collaboration with a front-end developer. Prototypes were created for both web and mobile platforms to validate interaction patterns and ensure consistency across devices before development began.

Validation Through User Testing: Feedback collected during the soft launch of both web and mobile applications informed multiple iterations. Usability issues identified during testing were addressed through refinements to layout, interactions, and visual hierarchy. Post-launch, Mixpanel analytics and session/video analysis were used to understand task completion time and repeated actions, guiding further improvements aligned with real operational workflows.