All projects · AI Agents & RAG
Savannah Autonomous AI Agent — Dynamic Buttons & Carousels
An autonomous website conversational AI agent ('Savannah') engineered with dynamic Generative UI capabilities.
An autonomous website conversational AI agent ('Savannah') engineered with dynamic Generative UI capabilities. Driven by prompt engineering ('Prompt your way through it') and built on the foundations of Puzzle #52, the agent converses naturally, answers complex questions, and dynamically produces functional interactive UI elements: responsive quick-action button matrices and multi-card visual carousels with live navigation telemetry, pagination dots, and lead booking CTAs. Supports Google Gemini, OpenAI GPT-4o, Groq, and an offline neural engine.
- Pixel-perfect replication of the challenge reference screenshot with dual-column split view (Left: Service Carousel, Right: Savannah Chat)
- Interactive 4-module Service Carousel with neon dashboard graphics, live index telemetry (1/4 through 4/4), and glowing cyan action button
- Full carousel controls: chevron arrows, 4-dot pagination cluster with direct jump, and keyboard navigation
- Responsive 7-button functional quick-action matrix matching screenshot labels: [BOOK NOW], [HOW CAN AI SAVE ME 10+ HOURS WEEKLY?], [SHOW ME SERVICES], etc.
- Generative UI engine on prompt: users can prompt the agent to dynamically create custom button clusters and multi-card carousels (e.g. pricing tiers, case studies)
- Prompt Engineering Studio tab featuring Savannah's system prompt directives and token grammar documentation
- Multi-LLM connectivity supporting Google Gemini, OpenAI, Groq, plus a 100% deterministic built-in offline neural engine
- Automated 8-point in-browser diagnostic test suite verifying navigation bounds, button clicks, and dynamic UI component generation
Built with: Conversational AI, Generative UI, Prompt Engineering, Interactive Carousels, Dynamic Buttons, State Machine, Google Gemini API, OpenAI GPT-4o, Groq API, JavaScript, Responsive Design