spill.

Answer a question
Step 1: Scan QR
Step 2: Share Feedback
Step 3: See Updates

Inspiration

We build spill because there are a lot of things we all feel we would do differently. From clumsy event planning to poor lighting in study areas, we want to talk about issues but no one bothers to raise them. Nobody had a low-friction way to report qualitative issues that actually led to change.

What it does

On our platform, users “spill” a thought (text, voice, photo), our AI hears the shape of the problem, stitches it to similar spills, and hands organisations clear, prioritized actions. Then the organisation posts short updates so people see outcomes. We celebrate fixes publicly so that people can see the power of effective feedback.

How we built it

We started by mapping every touchpoint where people already pause: lift lobbies, laundry rooms and lounges. We put QR-codes there and made targeted one-line questions in our React app for each of these areas. On the backend we trained an spacy NLP pipeline to turn messy human language into “what to do next.”

The goal of the entire build process was to minimize friction as much as possible. Thus, we worked hard to minimize app latency. We engineered query batching, deformalized reads and client-side caching to make server requests as efficient as possible. TRPC batching was key to this.

Challenges we ran into

The hardest thing wasn’t engineering. It was designing the UX to make feedback feel safe, immediate, and meaningful. We rewrote copy dozens of time so that we could convince users to try something new and to trust us.

If we learnt anything it is how hard it is to design a brand and interaction model that makes people stop treating feedback like a chore.

Accomplishments that we're proud of

Putting everything together into one package was our biggest achievement. We simplified collection into a single stream (reducing fragmentation), built an NLP stack to extract themes and action items, and prototyped timely update cards that increased re-engagement during trials.

Overall, building a product that is easy to interact with and genuinely exciting to use was our biggest win.

What's next for spill.

We want to implement this solution in the RCs to help them understand the pain points of residents. Then, we would like to expend across the campus to allow student voices to shape how culture and academics in NUS works. In the long run our goal is to give society a method collect and aggregate individual voices and concerns.

In the long run, where solving problems is becoming easier due to AI, we believe that understanding the human problems that need to be solved is of paramount importance.