Inspiration
The youth vaping epidemic is a global crisis, but we realized that current public health efforts are fundamentally misaligned. They deliver a single, generic message to a highly diverse, global audience. The crisis differs drastically by region, from black-market devices in Australia amid new regulations where local tobacconists are not allowed to sell vapes anymore, to highly addictive flavored pods in the US, to the unique challenges of disposable vapes in the UK. We were inspired to create No Filter to provide a dynamic, empathetic, and culturally precise public health intelligence platform tailored to local realities.
What it does
No Filter is a localized public health platform designed to help teens navigate the risks of vaping while empowering parents with actionable support tools. Upon selecting their region and role, users enter a tailored dashboard experience. Teens can access a strictly guardrailed AI chatbot for instant, region-aware advice, participate in interactive learning modules, or take educational quizzes. We also offer options to toggle between different languages. Parents utilize a dedicated "Family Co-Play" section designed for a shared-screen experience, bridging the communication gap through interactive, multiplayer learning. Localized health data, legal status, and cultural context are integrated into every interaction.
How we built it
We deployed a high-performance, cost-optimized tech stack that scales personalization efficiently.
Frontend: Built using React, Tailwind CSS, and Framer Motion to ensure a polished, accessible, and responsive user interface with dynamic conditional rendering.
Backend & API Orchestration: We utilized n8n Cloud for visual backend orchestration. We engineered a routing system that captures the user's region and dynamically injects localized static JSON health context before querying the AI.
Intelligence Layer: The application is powered by the Google Gemini 2.5 Flash API. We developed robust "System Guardrails" within the prompt logic to force the model to maintain strict public health focus, extreme brevity, and factual accuracy.
Infrastructure: Hosted via Netlify for the frontend client and n8n Cloud Production Webhooks for continuous, automated serverless operation.
Challenges we ran into
Transitioning our prototype to a live production environment presented significant technical hurdles. We faced the sudden deprecation of the Gemini 1.5 model, requiring an immediate migration of our API payloads to the Gemini 2.5 Flash model mid-development. Furthermore, we had to debug complex webhook lifecycle issues in n8n, resolving a critical blocker where the application was trapped in a single-execution testing loop rather than a persistent production state. Finally, engineering the AI to consistently return tightly constrained, formatted text rather than verbose or irrelevant answers (such as writing code) required extensive iteration on our system guardrails.
Accomplishments that we're proud of
We successfully engineered a bespoke Dynamic Regional Context Engine. This unique architecture delivers enterprise-level personalization for four distinct global regions with zero database hosting costs and zero API fine-tuning costs. We are immensely proud of developing an AI implementation that is highly localized, strictly guardrailed for public health, and capable of actively refusing off-topic queries. We proved that smart backend orchestration can rival the efficacy of custom-trained LLMs.
What we learned
This project was a masterclass in rapid, production-ready full-stack AI deployment. We gained deep insights into designing robust prompt guardrails for large language models to enforce strict persona compliance. Additionally, we mastered advanced backend orchestration with n8n, learning how to manage live production webhooks, state variables, and complex routing logic to securely manipulate data payloads before they interact with an AI model.
What's next for No Filter
-Our immediate roadmap focuses on expanding our technical capabilities and real-world impact.
-Vape Scanner Integration: Deploying our computer vision module to allow users to identify unknown vaping devices instantly via image recognition.
-Verified Community Directory: Launching a localized networking hub to connect parents with trusted professional resources and local support groups.
-Educational Pilot Programs: Partnering with educational institutions in Australian high schools to deploy localized versions of the platform in key school settings.
-Moderated Discussion Hubs: Building secure, region-specific community spaces for safe peer-to-peer support and engagement.
Built With
- gemini
- lovable
- n8n
Log in or sign up for Devpost to join the conversation.