Inspiration
While brainstorming, a problem came up that every one of us related to — choosing the wrong outfit at the wrong time and not making the most of the clothes we already own. It gets unexpectedly cold or snowy in the evenings, but you dressed for the morning and had no way of knowing. We all felt that, and we wanted to solve it. We also wanted to cut down the time and mental energy spent picking an outfit every single day.
What it does
Most mornings, we pick an outfit without checking the weather, without remembering what we wore last week, and without thinking about what the day actually demands. DayAdapt fixes that. You upload your wardrobe once, and every morning the app tells you exactly what to wear — based on today's forecast, your plans for the day, and what you haven't worn in a while. No more wet feet from the wrong shoes, no more reaching for the same three outfits, no more standing in front of a full wardrobe thinking "I have nothing to wear." It takes ten seconds and removes a decision you have to make every single day.
How we built it
We built the frontend with React.js and the backend with Node.js, using Firebase as our database and the Claude API as our LLM. The team split into two parallel workstreams — frontend and backend — coordinating through a shared API spec we defined upfront. This let us move fast and integrate cleanly right at the end.
Challenges we ran into
Our first challenge was scoping — it took real time to decide what we would and wouldn't build to keep things feasible within the hackathon window. The second was integration: connecting the frontend and backend surfaced a number of small but time-consuming edge cases we hadn't anticipated. The biggest time investment, though, was fine-tuning our recommendation system prompt to consistently produce high-quality, useful outfit suggestions.
Accomplishments we're proud of
We shipped a working v1.0 that actually does what it promises — taking a user's real wardrobe and returning genuinely good outfit recommendations. Getting the quality bar there within a hackathon timeframe was the thing we're most proud of.
What we learned
Beyond the technical skills, this hackathon taught us a lot about scoping, prioritisation, and working under pressure as a team. We also came away with a much deeper practical understanding of prompt engineering and how much it matters for LLM-powered features.
What's next for DayAdapt
DayAdapt has meaningful potential beyond v1. As users interact with the app over time, we accumulate rich data on preferences, wear patterns, and feedback — which could be used to fine-tune a specialised model for even more personalised recommendations. There's also room to expand into a weekly outfit planner, a laundry load tracker, and deeper health-aware features like UV protection nudges and activity-based outfit matching.
Log in or sign up for Devpost to join the conversation.