Inspiration
I’ve always been fascinated by how small risks can quietly escalate into bigger problems, whether in universities, campuses, workplaces, or public spaces. I wanted to create something that could reason through complex situations, identify risks, and explain their potential evolution. When I learned about Gemini 3 Pro’s reasoning and multimodal capabilities, I realized it could be the perfect brain for this idea. That’s how Sentinel AI came to life a system that doesn’t just detect hazards but thinks about them like a human expert, powered by cutting-edge AI.
What it does
Sentinel AI is a general-purpose risk reasoning platform that extensively leverages Gemini 3 Pro’s advanced capabilities: Multimodal reasoning it can analyze both text descriptions and images to detect potential hazards. Complex causal analysis it identifies how small issues can escalate into bigger risks over time. Explainable reasoning it shows not just what’s risky, but why it’s risky. Users can input a scenario from a campus lab, university building, workplace, or public facility, and Sentinel AI will generate risk assessments, escalation pathways, and early warning signals. It essentially functions as an AI safety companion that helps institutions anticipate and prevent accidents before they happen.
How we built it
I built Sentinel AI on Gemini 3 via AI Studio, fully utilizing its: High reasoning depth to simulate risk escalation in any environment. Multimodal input capability, allowing analysis of images and text together. Real-time interactive features, like rotating query suggestions and first-time onboarding. I focused on making it mobile-first and aesthetically pleasing. Additional features include: Why This Risk? Gemini 3 explains the reasoning behind each identified risk. Escalation Simulator it visualizes risk progression from immediate hazards to worst-case outcomes. History & Tracking keeping all user queries and outputs organized for easy review. This combination of Gemini 3 intelligence and thoughtful design makes it applicable across E.H.S sectors, universities, campuses, offices, or any place where risk understanding matters.
Challenges we ran into
Balancing domain flexibility while maintaining depth was tricky. We wanted Sentinel AI to handle all types of risks, not just E.H.S-specific ones. Getting Gemini 3 to reason correctly across text and image inputs required careful prompt engineering. Another challenge was ensuring mobile-first responsiveness, clean layouts, and intuitive UX without compromising analytical power. Finally, making sure the AI’s reasoning was clear and explainable to non-experts took multiple iterations.
Accomplishments that we're proud of
I’m proud that Sentinel AI is not just a tool it’s an intelligent assistant powered by Gemini 3 that can reason like a human expert. It handles multimodal input, risk escalation simulations, and explanations, all in a user-friendly interface. I’m especially proud of its applicability across sectors, from campus labs to workplaces, making AI-driven risk understanding accessible and actionable.
What we learned
We learned how to harness Gemini 3 Pro’s reasoning and multimodal features to tackle complex real-world problems. We also learned how important UX is even the smartest AI needs to present insights clearly and interactively. Most importantly, we learned how to translate AI’s complex reasoning into human-friendly guidance, which makes Sentinel AI truly valuable in H.S.E and general risk contexts.
What's next for Sentinel AI
Next, I want to make Sentinel AI become a robot that will automatically in real-time observe and detect risk and potential incidents and then give signal, also to expand interactive simulations, like predictive “what-if” scenarios using Gemini 3 reasoning. I’d also like to integrate automated risk scoring and comparative scenario analysis, making Sentinel AI even more actionable in workplaces, universities, or public facilities. Ultimately, the goal is for Sentinel AI to be the go-to platform for anticipating and managing risk anywhere, powered by the intelligence of Gemini 3.
Built With
- ai
- ai-studio
- gemini
- gemini-3-api
- html5
- react
- tailwindcss
- typescript
Log in or sign up for Devpost to join the conversation.