Exactly one year ago, on May 2, 2025, my co-founder Shishir and I did something terrifyingly raw.
We shoved a completely naked prototype onto a Netlify Drop build and pointed it at app.korture.ai. We did not launch it to the public. We just sent the link to a closed circle of friends and early testers to see if they could break it.
The data went straight into simple, raw form submissions routed through a backend file labeled 'korture-results'.
We built that initial experiment, called Korture ID, because standard resumes are completely broken. A resume is just a list of past titles and places where someone used to sit. It tells you nothing about how actual humans work together.
We wanted to see if we could map out a real professional blueprint using established behavioral science. The assessment took about 15 to 20 minutes and used two scientific layers under the hood:
Part 1: The O*NET Interest Profiler (60 targeted questions evaluating work interests using the RIASEC model).
Part 2: Work Style Profiling (15 questions tracking operational preferences based on the Job Demands-Resources model and psychological safety indicators).
Our goal was to prove we could map three distinct pillars:
→ Your natural day-to-day work style profile
→ The exact workplace culture where you feel most energized
→ The specific peer archetypes that naturally balance your strengths
We expected some polite praise from our friends. What we actually got was a wave of beautiful, confusing chaos.
When the results started hitting the 'korture-results' bucket, the feedback data was all over the place.
Some testers looked at their profiles and told us the results were spooky and accurate. But a lot of people did not play nice.
We had early testers get deeply offended by the data. They started furious, late-night arguments with us about why their operational profiles were wrong.
Another group had absolutely no clue what to do with the insights. They looked at the text, shrugged, and copied the raw data into early AI agents to use as custom prompts. Honestly, looking back at where AI models were in 2025, that was a brilliant piece of accidental user hacking.
It was messy, sarcastic, and loud. And it was the exact moment Shishir and I realized we were on to something huge.
Here is the thing: nobody gets into an intense, passionate argument over a standard resume. Nobody tries to hack a boring text document to feed it into an AI agent.
The fact that our early users cared enough to argue with the data or manipulate the tool proved that our data signals were hitting a real nerve. They were real, they were accurate, and they mattered.
That chaotic wave of mixed feedback gave us the exact conviction we needed. It proved we were not just building a fun side project. We were looking at a real solution to a massive corporate problem.
That scrappy Netlify build from a year ago laid the concrete for the platform we are scaling at Korture.com today. We are still building, and we are definitely still learning.
Your turn: When was the last time you launched an experiment that people completely hacked or argued with? What did you learn from the chaos?