Dante
Garetto

Work

  • UX Design Consultant · May 2026–July 2026 · Remote / On-site

  • Merchandising Intern — Hard Home Department · May 2025–July 2025 · Corporate HQ, Grand Rapids, MI

  • Mintify

    President, Project Director · August 2025–present · Ann Arbor, MI

    • Florida Community Innovation

      Mintify partnered with Florida Community Innovation to study how case managers and outreach workers find, verify, and trust information on the Florida Resource Map — a civic-tech platform combining a RAG pipeline, social-media monitoring, and NLP to surface real-time information on food pantries, shelters, job assistance, and healthcare.

      FCI wanted to know whether AI could help with two persistent problems: keeping listings current and matching people to resources that fit a specific need — a pet-friendly shelter, LGBTQ+ youth services, language-specific support.

      I directed the research team through three semi-structured interviews, 30–60 minutes each, with Florida-based nonprofit staff in outreach and case-management roles. Responses were affinity-mapped into five themes:

      • Geographic and accessibility barriers. Commutes of 45 minutes or longer, and distance filtering that returned results 25 miles out for a three-mile search
      • Information relevance and uncertainty. Mission statements where practical details should be, loose or misleading categories, and vague descriptions that discourage anyone from calling
      • The time cost of searching. Hours spent locating a single resource for a single family, compounded by hold times, phone trees, and verification done by hand
      • Trust that a listing is real and current. Providers re-check entries themselves, because they cannot assume an old one still stands
      • A cautious appetite for AI. Genuine demand for AI-driven efficiency, conditional on the interface staying simple and the underlying data staying reliable

      In their own words:

      I put in biking distance up to 3 miles, and I'm still getting more than 30 results — some of them are literally 25 miles away from me.

      They're putting on their whole story and who their board is … when you're in the day-to-day function of case management, you don't care who's on their board. You want to get … food.

      I'm spending three to four hours trying to find one resource for this family, versus just being able to upload and say, "This is what I'm looking for."

      Those findings became five recommendations, each answering a theme rather than a hunch:

      • Synonym-aware search. Recognize and group the different words people use for the same service, so finding a resource does not depend on guessing the term the directory happens to use
      • Primary and secondary tags. Structured tagging with a clear hierarchy, so a category cannot send a case manager down a path that wastes their afternoon
      • A guided search path. Simple prompts instead of an open-ended query box, for anyone who is unsure how to begin or has too little time to iterate on keywords
      • Multilingual support. Named directly by participants describing the Haitian and Brazilian communities they serve
      • Relevance-based ranking. Surface what is genuinely nearby, currently available, and matched to the stated need, rather than everything that matched a string

      The team then built those recommendations into a set of conceptual wireframes for the FRM, delivered to FCI with the final report.

        The FRM Data homepage: a blue header, a large ‘Welcome to FRM Data’ heading, a single wide search field prompting for either a website address or a description of the resource being looked for, a ‘Step by Step Research’ button beneath it, and two cards explaining how to use each of the two tools.

        A Flexible Way In

        1/2
        2025–2026

        Two entry points, because participants did not all arrive knowing what to ask for. The homepage takes a need in plain language, so finding a resource does not depend on guessing the directory’s own vocabulary, and ‘Step by Step Research’ is there for anyone who would rather be asked than have to phrase it. The Add Resources screen standardizes how a listing goes in — the ‘Additional details’ field captures the variations in how a service gets described, which is what makes it findable later.

        Swipe to browse

        Flexible Search

        A progress modal over the dimmed Add Resources screen, reading ‘Running…’ with 60% and ‘30 minutes left’ in green, above a horizontal progress bar.

        Feedback During the Wait

        1/3
        2025–2026

        Participants described searches that ran for hours with no sense of whether they were working. The progress state names a percentage and a time remaining, which bounds the wait and makes abandoning it less likely; the completion state says the run finished and offers the two next steps rather than leaving someone to figure out what changed; the error state says what failed and offers a way to look at why. All three are deliberately bare — the research found information overload was itself a barrier, so a status message is the wrong place to add more of it.

        Swipe to browse

        Progress and Outcome States

      • A results screen in three panels: ‘Name of the Organization’ with a list of its services, a ‘Service Details’ panel listing hours of operation, contact information, address, and accessibility; and a full-width ‘Credible Report’ block beneath them.

        The Essentials, Up Front

        2025–2026

        The detail view leads with what actually decides whether a resource is usable — hours, contact, address, accessibility — held in their own panel, separate from the organization’s name and service list, with a credibility report given a block of its own. It is the direct answer to the participant who described sites that open with their board of directors when what you need is to find out whether they have food today.

        Resource Detail

      • A color specification sheet headed ‘Colors,’ with four labeled groups — primary, secondary, tertiary, and neutral — each showing rounded swatches above a note on what the group is used for.

        Color System

        2025–2026

        The color set the wireframes are built on, each group labeled with the job it does rather than left as a bare palette: primary for calls to action and active states, secondary for sub-links, filters, and tabs, tertiary for feedback and warnings, and a neutral set for text, icons, and surfaces.

        Color System

      Skills

      • UX Research
      • Semi-Structured Interviews
      • Affinity Mapping
      • Thematic Analysis
      • Design Recommendations
      • Wireframing
      • Interaction Design
      • Information Architecture
      • Civic Tech
      • AI/RAG Products