About the Project
We built Arnabela to solve a simple but expensive problem: businesses often launch ideas without knowing how customers will respond. We wanted to create a way to test an idea before putting real money, time, and resources behind it.
Arnabela uses AI to simulate 100 distinct customer perspectives, each with different priorities, behaviours, risk tolerance, price sensitivity, and motivations. Businesses can enter a product, advertisement, offer, or concept and receive individual customer decisions alongside aggregated insights showing what drives interest, hesitation, and rejection.
What makes Arnabela different is that it doesn't stop at telling businesses what customers might think. It recommends a specific change and allows the business to create and test a second version against the same 100 simulated customers. This creates a simple experiment loop:
Test → Understand → Change → Retest → Compare
We learned that generating 100 AI responses alone is not enough. The customers need to behave differently and consistently. We therefore created 100 structured customer profiles across 10 behavioural archetypes, combined with an AI evaluation engine, simulation system, decision-driver analysis, and experiment comparison workflow.
We built Arnabela using Python, FastAPI, Pydantic, Google Gemini, Next.js, React, TypeScript, and Tailwind CSS. We also focused on reliability by validating customer data, handling failed evaluations and retries, tracking simulation progress, and ensuring Version 1 remains unchanged when creating Version 2.
One of our biggest challenges was working with limited AI API availability while building a system designed around 100 evaluations. To overcome this, we created a deterministic mock evaluation mode that mirrors the real evaluation structure, allowing us to test the complete platform without continuously consuming API quota.
The result is a repeatable way for businesses to learn before launching:
100 different perspectives → 100 different decisions → one clearer direction.
Our goal is simple: help businesses test their ideas before the market does.
Built With
- a/b-testing
- automation
- business-intelligence
- consumer-insights
- customer-simulation
- data-analytics
- decision-support
- experimentation
- fastapi
- javascript
- market-research
- next.js
- product-testing
- pydantic
- python
- react
- rest-api
- synthetic-customers
- tailwind-css
- typescript
Log in or sign up for Devpost to join the conversation.