Crafted bouquets, guided by AI
SUMMARY
FlorAI, a mobile app that uploads an inspo photo, identifies the flower types and hex palette. Makes material checklist, build-steps, and completion post. Solo app development project.
PROBLEM STATEMENT
The knowledge barrier to arranging bouquets is high. Users either overpay for overpriced pre-made bouquets or hesitate to DIY, creating stress when they want to provide a personable gift, for any occasion.
SUPPLEMENTAL
ROLE(S)
Design
Development
CONTEXT
Personal Project, AI Assisted
TIMELINE
May 25' - Current
PLATFORMS
Mobile, iOS Build
TEAM
Zander Vilaysane
PROCESS
( 1 ) User Research
Conducted user interviews with 8 participants (low-med experience) to follow storyboard of making arrangements and observing their points of priority and frustrations.
Deployed survey with 56 responses to observe interactions with general crafts (not just floristry).
Analyzed adjacent platforms (Pinterest, bouquet tutorial blogs, food recipe apps) to benchmark FlorAI’s necessary features vs. nice-to-haves.
( 2 ) Insights
Interviews displayed a general user flow of: finding inspo on social, buying bundles at stores, assembling at home. Entry florists spent most of their duration at the store looking up floral details. Medium florists spent most time at assembly, focused more on experimentation.
Surveys indicate only 16% of users do crafts in formal studio spaces, while 83% prefer to do crafts at home or in social gatherings.
Arrangement process is split between inspiration-only (Pinterest) and instruction-heavy narrative-dense blogs. Meanwhile, recipe apps showcase successful components: clear ingredient lists, step-by-step instructions, and repeatable outcomes—but they’re for food, not flowers.
( 3 ) Solution
Recenters main feature to be intended for entry florists to advance to medium skill level through confidence building and practice. Emphasizes the learning factor of useful inspo extraction and detail display.
Features should provide mentorship and guidance that can be followed individually and minimize dependence on external research tools.
Applying proven food recipe pattern to arrangement process alongside competitor strengths. Inspo image -> palette, stem counts, flower type -> material checklist -> guided build steps.
( 4 ) Testing & Iteration
Tested Vision API extractions through different prompting approaches to extract different data fields / accuracies of flowers in photo. Iterated through instruction phrasing templates to find most general, but visually understood rules. Prompting for better, more curated instructions is a work in progress.
Beta testers explored on Expo Go and TestFlight responded with their own upload inspirations. Validated accuracy consistency and provided notes of where instruction steps could be reformatted.
Concepts of Find Nearby Stores, Price Budgeting, Further Personalization were prototyped as nice-to-have considerations. Usability testing found them useful in itself, but required cleaner, more intuitive iterations.
OUTCOME
API response consistently outputs average confidence rates of > 70% on flower identifications and color palettes. > 60% for stem counts.
Refined project overview details (captioning, tags) and curated floral tips (processing, care) to improve personalization feel to experience.
Seamless, decluttered flow of upload -> identified list + add-on suggestions -> checklist -> instruction steps + overview of steps + curated floral tips -> project overview
ENDING REMARKS
Optimizing Prompting Approach
API integrations provide many utilities to improving the user experience. To get desirable outputs from these packages, it's essential to understand prompting mechanisms and approach.
Impacts of Automating the Meticulous Parts
Users spent less time researching (streamlined by API output of detail display of IDs and curated tips) and focused more on creating with higher quality and creative vision. FlorAI became the baseline for experimentation; confidence and willingness to try new styles increased. Made arrangements more frequently and for more diverse occasions.
Allowing for Scalability
As ambitious as I was to develop more and more features, it was more controlled to ship a MVP and incrementally update with desirable features. Understood value of beta testing to conduct feature exploration/feasibility.
visual artifacts
GENERAL FLOW (TOP), INTERNAL COMPONENTS FLOW (BOTTOM)
User Flow - Main Feature
