Building Futures,
One Block At A Time

The Vision

Mindo, Ecuador, is a world-renowned cloud forest ecosystem and a global capital for biodiversity. However, the digital infrastructure for tourists navigating this sanctuary was fragmented. The goal was to build a centralized, highly interactive platform that connected international travelers with local eco-lodges, gastronomy, and adventure activities, while simultaneously serving as a dynamic, community-driven map of the forest itself.

Phat Stack Technologies deployed a next-generation smart tourism portal. We moved beyond a standard static directory, utilizing advanced AI and geospatial mapping to create a living, breathing digital guide for the Mindo valley.

The Challenge

• Language Barriers: Mindo attracts a truly global audience, meaning a simple English/Spanish toggle was insufficient.
• Information Silos: Local businesses, hidden waterfalls, and trails were largely documented through word-of-mouth or outdated maps.
• Complex Data Routing: The platform needed to cleanly categorize distinct verticals: Real Estate, Eco-Lodges, Restaurants, and High-Adrenaline Activities.

Technical Architecture

To achieve the speed, AI integration, and complex mapping required for Mindo.info, the project utilizes a modern, decoupled tech stack designed for scale.

• Frontend Architecture: A modern component-based framework (like Next.js or React) to ensure fast client-side routing, optimal SEO for tourism keywords, and seamless state management when users toggle between the map and directory listings. Styling uses Utility-first CSS for rapid, responsive design that adapts flawlessly from desktop planning to on-the-go mobile viewing. Geospatial UI integration with an interactive mapping library (such as Mapbox GL JS or Google Maps API) handles GeoJSON data for clustering nature spots.
• Backend & Database: A RESTful or GraphQL API environment handles the heavy lifting between user requests, the database, and external AI models. A relational database (PostgreSQL) manages complex relationships. Secure JWT-based user authentication handles traveler profiles.
• AI & Vision Integrations: Conversational AI powered by an LLM serves as the "Expert AI Guide," prompt-engineered with specific local context. A Computer Vision Pipeline for the Nature Spotter classifies biological entities from uploaded photos, extracting EXIF data for coordinate mapping.
• Deployment & Infrastructure: CDNs serve heavy media assets and execute edge-routing to detect user locales and deliver the correct language instantly.

The Outcome

Mindo.info stands as a premier example of how AI and modern web architecture can elevate local eco-tourism. It successfully bridges the gap between celebrating a tight-knit local community and providing a frictionless, high-tech experience for global travelers.

The Challenge

Presenting advanced biophoton technology and energy healing services in a way that is immediately understandable and accessible, especially in fast-paced live event environments.

The Solution

We designed the platform to function like a digital kiosk. The homepage uses clear, tappable cards that allow visitors to instantly choose products or sessions, streamlining the learning and purchasing process via integrated affiliate links.

The Outcome

A streamlined visual gateway that effectively educates potential customers and drives conversions on the official Eluumis and Biohealing Stream platforms.

The Challenge

Starflower Consulting needed a brand identity that communicated professionalism without feeling stiff or overly corporate. The goal was to stand out with an engaging, vibrant aesthetic.

The Solution

We developed a "Playful Professional" visual identity. By taking static branding elements and introducing fluid motion design, we created an dynamic experience that instantly captures attention while maintaining trust.

The Outcome

A memorable, modern brand presence that effectively conveys the consultancy's unique energy and approach.

The Architecture

The application is built completely serverless using FastAPI on Google App Engine, backed by Google Cloud Datastore for rapid NoSQL transactions. It deeply integrates with Google's Gemini AI to automate customer service workflows at scale.

Omnichannel & AI Automation

The system ingests tickets from a custom Alpine.js stateful web chat, direct Email synchronization, and Telegram bots. The Gemini engine performs autonomous triage—determining sentiment, assigning risk profiles, and generating draft responses for human review based on ingested PDFs and web crawls.

Safety & Administration

Features a comprehensive stats dashboard for performance metrics and real-time AI cost tracking. An automated Budget Safety Net monitors API costs and triggers maintenance mode if budgets are exceeded. Automated Google Cloud Scheduler jobs handle routine cleanups and backups.

The Vision

Biohealing Stream visitors often arrive knowing they want support — but unsure which Living Light protocol fits their situation. The goal was a fast, welcoming assessment that honors each visitor's stated primary concern, recommends the right protocol with confidence, and hands them off to Biohealing Stream with a clear next step — while giving operators the data to see who's arriving, where they drop off, and which protocols win.

The Challenge

• Complex Scoring Logic — 15 questions map across the full Living Light protocol library, including multi-select body-system questions and weighted primary-focus anchoring across Mind and Body categories.
• Conversion Without Friction — Premium feel, under one minute, iframe-embeddable across the Biohealing ecosystem, with all scoring logic kept strictly server-side.
• Operator Visibility — Funnel drop-off, UTM attribution, completion timing, and version-stamped session data had to be captured from day one without adding any friction to the user experience.

The Architecture

Built fully serverless on Google App Engine (Python 3.11) with FastAPI and a glass-morphism single-page frontend. The backend-only scoring engine processes the full protocol library through weighted Q1 anchoring, priority tiebreakers, and de-duplicated physical-system scoring — the logic is never exposed to the client. Session tracking opens at quiz start, per-question progress is recorded via sendBeacon, and completed submissions persist to Google Cloud Datastore. GitHub Actions handles continuous deployment on every push to main.

Scoring & Protocol Matching

The engine anchors each recommendation to the visitor's stated primary concern before weighing any secondary signals. Multi-select body-system questions feed contextual scoring that de-duplicates overlapping physical signals. Every protocol competes through priority tiebreakers that prevent ambiguous ties, and the full engine is covered by automated tests so output is reproducible and auditable.

Analytics & The Garden Dashboard

Each session captures UTM source, device type, and browser fingerprint at quiz start. sendBeacon fires on every question advance, producing a per-question drop-off map that survives page closes. The password-protected /garden dashboard surfaces protocol win rates, answer-level distributions, device and UTM breakdowns, and completion funnels — all filterable by Eastern-time date range with no external analytics dependency.

Frontend Experience

Zero-build stack — vanilla JS with Tailwind CSS. Animated question cards transition smoothly between all 15 questions. Result pages render protocol-specific emblem artwork, anatomical diagrams, and a score visualization before presenting a direct conversion CTA to Biohealing Stream. The entire quiz is iframe-embeddable with CORS configured for cross-domain deployment at quiz.biohealing.stream.

The Outcome

Visitors move from a branded welcome screen through 15 animated questions to a rich, protocol-specific result page — complete with custom artwork, score visualization, and a clear next step. Operators see exactly where users abandon and which protocols win. A data-backed onboarding funnel with zero server maintenance and a deployment pipeline that ships on every git push.

The Architecture

The plugin is built using modern, object-oriented PHP principles for WordPress. It features a modular, single-responsibility design initialized via a Singleton Core class, ensuring scalability and maintainability while loading heavy modules only when properly licensed.

Agentic Generation & Strict SEO

Leveraging Google's Gemini AI, the system uses "Deep Thinking SEO Prompts" to mandate strictly formatted JSON responses. It guarantees 100% Yoast scores by strategically placing keyphrases, injecting links, enforcing readability rules (banning passive voice), and generating meta descriptions.

Advanced Native Integration

Operating asynchronously via WordPress AJAX hooks, the interface allows bulk-queuing without page reloads. The backend schedules background processing via WooCommerce ActionScheduler with business hour simulations and randomized jitter, while programmatically sideloading AI-generated images with complete SEO metadata directly into the WordPress Media Library.