Veronnik

Inspiration

We’ve all experienced it: facing hundreds of sports jackets online, wondering: "Is this actually waterproof enough for my hike tomorrow?" or "Will I receive it before my bike trip this weekend?" In the sports industry, 70% of carts are abandoned because of technical doubt. We wanted to bridge the gap between the expert advice you get in a physical store and the convenience of e-commerce.

What it does

Veronnik is a proactive "Contextual Shopping Agent" for sports retailers. Unlike a basic chatbot, it bridges three real-time data streams:

User Intent: What the user is looking at.
Environmental Context: Live weather and geolocation (e.g., "It's going to rain in Paris tomorrow").
Operational Reality: Real-time local store stocks. It greets the user with a proactive advice: "I see you're looking at this bike; since a storm is coming to your area tomorrow, here are the 3 best mudguards currently in stock at your local store." ## How we built it We built a full-stack MVP in record time using a modern low-code/AI stack: • Frontend: Lovable (React, Vite, and Tailwind CSS) to create a seamless UI that mimics a native integration on a major retailer's site (like Decathlon). • Intelligence: Featherless.ai to power our LLM, allowing for expert-level reasoning on technical gear. • Database: Supabase to manage our real-time product catalog, weather tags, and stock levels. • Architecture: We designed it as a "Plug-and-Play" widget that can be injected into any existing e-commerce site with a single line of code ## Challenges we ran into The biggest challenge was Contextual Filtering. It’s one thing to show products; it’s another to make the AI "understand" why a specific weather condition (like a 10°C drizzle) requires a specific technical fabric (Gore-Tex vs. Softshell). We had to fine-tune our System Prompts and structure our database tags carefully to ensure the recommendations were always safe and relevant ## Accomplishments that we're proud of We are happy about our team working experience. ## What we learned Learning some new tools for the project and insert lovable and featherless.ai. ## What's next for Veronnik We believe that interactive kiosk we Veronnik could be set up in store for users. We could also use Veronnik for different models, use it as a cross over model for different problematics.