Building local & secure survey-analysis tools and other recent examples of using AI in design work

Some recent, possibly inspiring examples of how I’ve found AI useful in my design and product work. The most interesting thing AI has changed is not speed. It’s what kind of things I can now make real enough to evaluate.

Tech validation. There were questions about whether React Native currently provides good enough gesture and animation support, support for native UI widgets, proper offline caching and background push. Instead of guessing or spending developer time on an expensive POC project, I built a test app with Claude in a matter of days to try it out for real.

Survey analysis. I needed to aggregate data from 12 different surveys to analyze it further. On top of that, I wasn’t allowed to share the data with external cloud tools.

I used Claude to analyze the CSV syntax that the survey tool produced, then asked it to build a local viewer app that reads the raw CSV files and lets you filter, pivot and visualize the data. You can also export the raw results in .md format to feed to your corporate AI for further analysis.

I wrapped the whole thing inside a single HTML page that’s easy to share with anyone who wants to dig into the data.

Tech validation. There were concerns about whether one feature idea could be built directly with native iOS map tools. Two hours later we were wiser – and had a working demo to help sell the idea internally.

Design exploration. The online store had no design files to build on. With AI, I created a replica of the live service as a starting point, then designed alternative new features and layout and visual variants on top of it, with the existing checkout flow staying functional underneath.

Concept validation. One startup had an ambitious concept that only makes sense if you can collect the data and pair it with a smart AI that actually uses it. I have no idea how I could have designed and prototyped that a year ago in Figma.

Now, a couple of days later, I have a working product draft with realistic mock data, powered by the Anthropic API for its AI features. I can test what the experience actually feels like, and we can design the product around this real, living core together with the founding team.

Let me know if you’d want something like this. I don’t have much time these days – but then again, not much is needed anymore.

Original post

Verkkotyökalu kyselytulosten analysointiin, jossa esitetään kaksi ideaa ja niiden arviointijakaumat tähdillä (1-5) sekä avointen kommenttien määrä.
Kaavio, joka näyttää hyödyllisyyden ja kiinnostuksen suhteen survey-tuloksista. Bubble-tyyppiset pisteet kuvaavat eri ideoiden arvosanoja ja niiden saamaa kiinnostusta.

Jätä kommentti