VESTA: The Proactive Daily Compass

💡 Inspiration

We stand on the edge of the "Silver Tsunami." By 2030, 1 in 6 people worldwide will be aged 60 or over. Society often views this demographic through a lens of decline—focusing on care, monitoring, and safety.

As a 61 year old, not log from retirement person myself, I asked a different question: What if I designed for vitality instead of decline?

My inspiration came from "The Connection Economy." I identified a dual crisis facing the modern elder:

1. A Crisis of Connection: Loneliness is as damaging to health as smoking 15 cigarettes a day.
2. A Crisis of Purpose: Decades of professional expertise go dormant upon retirement, creating an 80% gap between seniors who want to volunteer and those who actually do.

I built VESTA to be the operating system for this new economy. It empowers seniors not just to stay safe, but to stay relevant, connected, and creative.

🚀 What it does

VESTA is a High-Agency platform designed for tablets and ambient smart displays. It replaces the cluttered app grid with a calm, intelligent interface centered on three pillars:

1. The Skill Bridge: Matches the user's lifelong expertise (e.g., Architecture) with local micro-volunteering opportunities (e.g., "Consult on a historic roof repair").
2. Generative App Creator: Allows users to build their own digital tools simply by describing them. "I need a log for my orchid watering" becomes a fully functional mini-app in seconds.
3. The Horizon (Grounded Planning): A daily planner that uses Google Search Grounding to check real-world logistics (tides, traffic, weather) before suggesting an itinerary.

⚙️ How I built it

VESTA is a production-grade web application built with Angular 19, leveraging its latest reactive primitives (Signals) and Zoneless architecture for maximum performance on low-power devices.

The brain of VESTA is Google Gemini, integrated via the Google GenAI SDK. I utilized a Service-Oriented Architecture to treat the LLM as a deterministic logic engine:

• Gemini 3 Flash Preview: I used this model for its speed and reasoning capabilities. It powers the "App Creator" and "Daily Insight," using strict JSON Response Schemas to ensure the AI generates valid data structures that the UI can render instantly without breaking.
• Google Search Grounding: For the "Horizon" planner, I enabled the googleSearch tool. This allows Gemini to fetch real-time data—like the exact time of high tide in Richmond or current traffic conditions on the M25—ensuring VESTA provides accurate, actionable advice.
• Zod Schema Validation: To bridge the gap between AI creativity and software stability, every AI output is parsed and validated against Zod schemas before reaching the application state.

🚧 Challenges we ran into

• Designing for Dignity: Most "senior tech" looks patronizing (huge buttons, primary colors). I struggled to balance high accessibility standards (contrast, font size) with a sophisticated, "magazine-like" aesthetic that respects the user's taste.
• The Hallucination Problem: In early tests, the planner would invent train schedules. Implementing Google Search Grounding was the breakthrough moment, turning VESTA from a creative writer into a reliable secretary.
• Structured Output Stability: Getting the "App Creator" to consistently generate valid SVG paths and Tailwind color classes required extensive prompt engineering and strict schema definitions.

🧠 What I learned

• AI as an OS Layer: I learned that LLMs shouldn't just be chatbots inside an app; they should be the app logic. VESTA doesn't just talk; it routes data, builds UIs, and manages state.
• The Hunger for Purpose: In user research (simulated via our persona, Harry), I found that tools facilitating contribution (volunteering, teaching) were far more engaging than tools for passive entertainment.

🔮 What's next for VESTA

• Hardware Integration: I envision VESTA running on dedicated ambient hardware (like a Google Nest Hub Max) with proactive voice presence.
• Expanded Skill Bridge: Partnering with platforms like VolunteerMatch to bring real-time API data into the Skill Bridge.
• Family Circles: Deepening connections to allow secure, private social sharing that doesn't rely on ad-supported social networks.

Built With

• ai-models:-gemini-3-flash-preview-(logic/reasoning)
• imagen-3-(image-generation)
• search
• signals).
• tailwind
• typescript
• veo-2-(video-generation).-ai-tools:-google-search-grounding.-frontend-framework:-angular-19-(zoneless
• zod.