Skip to content

Product Explosion Report: The Operating System for Outdoor Intelligence

[!CAUTION] The Brutal Truth: What We’ve Wasted Time Building We have been building a Vite application with an AI chat sidebar. We have treated the map as a static component and the AI as a conversation thread. We have been manually stringing together UI, state, and offline caching when entire foundations exist to solve this. If we continue on this path, we will own a fragile, custom-built application that requires constant maintenance instead of a robust intelligence platform.

[!IMPORTANT]
Core Engineering Principle
BluGuide owns fishing intelligence. The open-source ecosystem owns mapping, rendering, synchronization, identity, communications, storage, and visualization. Every architectural decision begins by identifying the smallest stable upstream owner before custom code is introduced.

1. The Vision: The OS for Outdoor Intelligence

BluGuide is not a fishing app. It is the first domain implementation of The Operating System for Outdoor Intelligence.

By building a domain-agnostic Context Engine, we immediately open the platform to: * BluGuide Maine * BluGuide Salt * BluGuide Hunting * BluGuide Hiking * BluGuide Snowmobile * BluGuide Forestry * BluGuide SAR (Search & Rescue)

Same platform. Different domain packs.

2. The Architecture (Context > Map)

The map is not the product. The map is merely one projection of the underlying World Model. 

graph TD
    subgraph BluGuide OS
        CE[Context Engine]

        CE --- K[Knowledge]
        CE --- LS[Live Signals]
        CE --- PH[Personal History]

        K --- DE
        LS --- DE
        PH --- DE

        DE[Decision Engine]
    end

    subgraph User Experience
        Voice
        Map
        Widgets
        Watch
        Dashboard
        BoatDisplay[Boat Display]
    end

    DE --> Voice
    DE --> Map
    DE --> Widgets
    DE --> Watch
    DE --> Dashboard
    DE --> BoatDisplay

The AI disappears. It is replaced by ambient intelligence (the Context Engine) that understands the world, reasons over it, and projects it to the appropriate interface.

3. The Open-Source Arsenal

We will eliminate custom infrastructure by aggressively adopting mature upstream projects.

Geospatial & Visualization (Open Source Styling)

  • OpenMobileMaps: We will adopt this instead of owning map styling.
  • MapLibre GL JS + Deck.gl (MIT): The rendering engine.
  • Geospatial Formats: PMTiles, MBTiles, GeoParquet, FlatGeobuf. We own the intelligence; we do not own the spatial formatting.
  • Turf.js (MIT): Client-side spatial analysis.

The Input Adapters (Signals)

Signal K is fantastic, but it is just one adapter. The Context Engine consumes telemetry from a vast array of inputs: * Vehicles/Vessels: Boat, Vehicle, Drone. * Marine Electronics: Garmin, Lowrance, Humminbird, AIS, NMEA2000, Signal K. * Wearables & Mobile: Watch, Phone, Bluetooth sensors. * Environment: Weather, USGS, NOAA.

Offline State & Synchronization

  • ElectricSQL vs. PowerSync: We will not choose one yet. We will perform a formal capability audit evaluating: conflict resolution, licensing, iOS support, SQLite extensions, Postgres compatibility, replication topology, offline writes, partial sync, and spatial support. The winner is whoever owns the larger percentage of the problem.

4. Capability Packs & Composition

The application becomes a composition of specialized packs, entirely aligning with the Gas City architecture:

  • Weather Pack
  • Bathymetry Pack
  • Species Pack
  • Navigation Pack
  • River Flow Pack
  • Boat Sensors Pack
  • Satellite Pack
  • Fish Identification Pack
  • Forecasting Pack
  • Trip Planning Pack
  • Emergency Pack
  • Offline Maps Pack

5. Knowledge as a First-Class Capability

Knowledge is the engine’s fuel. Everything becomes one interconnected semantic graph:

  • Every lake
  • Every river
  • Every regulation
  • Every stocking report
  • Every guide article
  • Every catch log
  • Every trip
  • Every sonar recording
  • Every photo
  • Every weather observation

The Context Engine queries this graph instantly. We will leverage tools like Orama for offline edge search and Qdrant for deep backend RAG, ensuring the semantic graph is always accessible, even in the woods without cell service.

6. Six Months From Now

When you open BluGuide on your phone in the boat: 1. You do not see a chat screen. You see your World Model projected onto the Map. 2. The intelligence is ambient. Because the Context Engine knows where you are, what time it is, and cross-references your Personal History with Live Signals (SunCalc, SQLite, NOAA), it proactively projects decisions: "Barometric pressure is dropping and the solunar major period begins in 15 minutes. Switch to copper and troll the western drop-off." 3. You own the domain, not the infrastructure. We have replaced 70% of our planned custom code by integrating these mature projects. Our engineering effort is spent exclusively on Bluefly’s outdoor intelligence.