Experiments powered by AI.
I use AI to build and test digital products fast. These projects help me explore how design, automation, and language models can turn simple ideas into working tools.
I combine years of product design experience with AI coding tools to prototype, ship, test, and improve products quickly. Each project is a way to learn by building.
AI Assistant
Bentevi
Bentevi is an AI chat assistant designed to be simple to train and easy to embed on any website. A futute version of this Ai mode I have on my portfolio.
I use ChatGPT to plan the roadmap and product iterations, OpenAI Codex to built the POC and Anthropic Claude
Code to build the prod version. The goal is to remove the complexity of traditional chatbots and agents,
creating
a lightweight and flexible AI assistant.
The project combines interface design, prompt design, token optimization, knowledge data, and user context
to deliver high-quality AI conversations while minimizing hallucinations.
Coffee Origin Tracker
Pindorama
Pindorama is a tool to check where specialty coffee comes from and if that origin can be trusted.
Users can search by farmer name or scan a QR code to see the farm on a map, basic origin data, and a simple
risk score based on location and deforestation signals.
The tool is focuses on Brazilian coffee
and
combines public data, satellite imagery, and user contributions. Over time, roasters and exporters can add
or verify information.
Users can also attach tasting notes to each origin, linking flavor to place, rate roasters and brands that
use the same coffee beans location and buy or compare the available products around them. The project
explores how to turn fragmented, technical data into
something simple and useful for everyday
coffee drinkers.
Health programs powered by AI
daylist.co
Daylist is a service that builds personalized health programs using AI. It uses customer data and biomarkers to suggest supplements, nutrition, vitamins, and wellness products. The goal is to support both short-term health and long-term outcomes.
AI tools are part of how the product is designed and built. They help with planning, iteration, and
development.
Because this involves health data, privacy and EU regulations are part of the foundation, not an
afterthought.
The service depends on partnerships with testing labs, healthcare providers, online pharmacies, and product
suppliers, including white-label manufacturers.
