Case study · Web platform · 2025
SefAIz: AI, gamification and accessibility to make tax tasks simpler
An MVP proposal for the Pernambuco State Finance Department (SEFAZ-PE) that teaches entrepreneurs how to get compliant and access tax incentive programs, with learning paths, an eligibility simulator and an AI assistant.
The problem
State incentive programs such as Prodepe can change a company’s reality. But the rules are dense, the language is technical, and the path to joining is full of unclear steps.
The result shows up on both sides of the counter: companies that give up or get their application wrong, and auditors who receive inconsistent cases and pile up rework.
Process
The project went through more than a month of immersion, with ideation and definition stages, client interviews and research into the problem. The format was inspired by AWS and Google Cloud learning labs: hands-on content in short paths with visible progress.
Besides designing, I was the bridge to the client, aligning expectations and turning their interests into an MVP scope.
The solution
Instead of yet another document portal, a platform that teaches by doing. Each path takes users from concept to practice, and AI shows up where it helps people learn, not as decoration.
Learning paths
Content split into short steps (introduction, simulator and quiz), with estimated time and progress.
Eligibility simulator
Companies check whether they qualify before putting an application together, cutting errors at the source.
AI as a learning tool
A chatbot with personalized suggestions, audio transcription for people with visual impairments, and the most relevant searches highlighted.
Gamification
Badges, certificates and a public profile to keep people engaged along the paths.
Design decisions
Assistive design
A label for the course in progress, a visual and numeric progress bar and the current page highlighted, so users never get lost in dense content.
User in focus
Colors and labels mark the course in progress, and the sidebar shows the features currently in use.
Accessibility from the MVP
Audio transcription was part of the initial scope, not a future improvement.
What came next
The proposal evolved into Fiscolab, a platform built as a team that SEFAZ-PE praised highly and that became a scientific paper presented at IHIET-AI 2026.
