melantha
AI platform that simulates human behaviour

Challenge
Melantha helps people solve personal dilemmas by showing what others would do. AI agents run a simulation in real time and calibrate against the person’s own answers. The product needs a simple interface, strong storytelling and reasons to come back.
User flow

Answering others’ dilemmas
Users need to give before they get. Medium friction, so I made the questions straightforward and the process simple and fast: under 30 seconds to answer.

Personal dilemma
The AI asks a few questions before giving an overview of what other people will see. A familiar interface with new things: the person can change the overview right in the chat, and 50 AI agents answer one by one, as an animation in the chat, which makes the answers feel real and builds trust.

Result
Two simple graphs right inside the chat, so users don’t leave the conversation to see results. Easy to understand, with potential to go viral.

Feedback
This window pops up when a user asks their second dilemma, a sign they’re getting value. It collects quantitative feedback alongside product analytics to understand user intentions.
What’s next
I started Melantha as part of my module, Design Innovation Project. It’s still super early, so first I need more user feedback and analytics. To get users, I need a good marketing channel, ideally viral ideas. The product is free and built for research. The metric I’m watching: how many users come back for a second dilemma.