Objectives of the service
Pelagic fishing faces a critical challenge: fish schools move unpredictably through vast oceans, forcing vessels to burn precious fuel searching blindly. Traditional knowledge fails as climate change disrupts migration patterns.
Pontos Prisma deploys an AI agent system powered by a specialized marine foundation model trained on millions of satellite observations, oceanographic data, and fishing patterns. This intelligent agent continuously analyses thermal fronts, chlorophyll concentrations, and current flows to understand where pelagic species congregate.
The system operates as multiple coordinated agents: a Prediction Agent that forecasts school movements 12 hours ahead, a Risk Assessment Agent that evaluates environmental conditions for sustainability, and a Fleet Coordination Agent that optimizes collective fishing strategies across vessels.
Fishers interact through natural voice commands with a Conversational Agent that understands maritime context in multiple languages. The foundation model learns from each fishing trip, improving predictions through reinforcement learning based on actual catch data.
For cooperatives, specialized agents handle market price optimization, compliance reporting, and resource allocation, transforming disparate boats into an intelligent fleet network.
This agent-based architecture ensures fishers find tuna, mackerel, mahi mahi, and many more efficiently while maintaining sustainable practices, reducing fuel consumption by 40% and increasing catch rates by 35%.
Users and their needs
Pontos Prisma targets pelagic fishing operations across the EU Atlantic coast, initially focusing on Portugal, Spain, and France where 35,000 vessels (5-25m) pursue highly mobile species like tuna, mackerel, and sardine.
Primary Users:
-
Individual Vessel Captains: Operating purse seiners and pelagic trawlers
-
Cooperative Managers: Coordinating 20-200 member vessels
-
Fleet Operations Managers: Optimizing 2-20 vessel companies
-
Crew Members: Using voice logging and safety features
User Needs:
-
Locate fast-moving pelagic schools while optimizing fuel
-
Predict species type before deployment (tuna worth 10x more than mackerel)
-
Coordinate with other vessels to avoid empty searches
-
Meet MSC sustainability certification requirements
-
Access real-time market prices for catch timing decisions
-
Simple interfaces that work with wet, cold hands
-
Offline functionality for areas without connectivity
-
Native language support (Spanish, French, English, Portuguese)
Key Challenges: Meeting these needs requires overcoming significant technical hurdles: achieving 80% prediction accuracy for highly mobile species, processing satellite data fast enough for real-time decisions, building trust among traditionally competitive fishers to share data, and creating interfaces simple enough for aging fishers (average age 52) yet powerful enough for modern operations. The harsh marine environment demands exceptional voice recognition and offline capabilities.
Service/ system concept
Service/System Concept
Pontos Prisma delivers hourly predictions showing where pelagic fish schools will be, which species to expect, and optimal fishing conditions.
Key Capabilities:
-
12-hour advance predictions with 80% accuracy
-
Species discrimination
-
Risk assessment for sustainable catch zones
-
Real-time market prices and quota status
-
Fleet coordination and intelligence sharing
-
Voice-controlled logging in 4 languages
-
Offline mode with automatic synchronization
How It Works: Every 90 minutes, Pontos ingests satellite data from Sentinel and Copernicus, detecting ocean thermal fronts and chlorophyll patterns where pelagic fish aggregate. Our AI foundation model, trained on millions of fishing events, analyses these conditions alongside current flows and historical patterns.
The system operates through three layers:
-
Space Layer: Satellites capture ocean conditions
-
AI Layer: Machine learning models predict fish movement and assess sustainability
-
Delivery Layer: Mobile apps provide captains with simple, actionable intelligence
The AI agents continuously learn from fishing results, improving predictions with each trip. Fleet data sharing multiplies intelligence across the cooperative, turning individual boats into a smart network. All processing happens in the cloud, delivering insights directly to fishers' smartphones even in rough seas.
Space Added Value
Pontos Prisma leverages multiple Earth observation satellites orchestrated through advanced AI to create comprehensive maritime intelligence:
Primary Space Assets:
-
Sentinel-3 OLCI & SLSTR: Detects chlorophyll patterns and thermal fronts where pelagic species naturally aggregate
-
Sentinel-1 SAR: All-weather monitoring of sea surface patterns and vessel movements
-
Jason-3/Sentinel-6: Ocean current dynamics affecting nutrient transport and fish migration
-
Copernicus Marine Service: Integrated oceanographic parameters for multi-layer analysis
AI Orchestration - The Unique Differentiator: Unlike competitors creating layers of oceanographic maps for users to analyse, Pontos' Deep Reinforcement Learning system orchestrates multiple data streams to provide recommendations:
-
Multi-Agent Optimization: Coordinates governance (quotas), economic (prices), social (fleet behaviour), and biological (fish migration) layers
-
Predictive Intelligence: Transforms fragmented satellite data into actionable 12-hour forecasts
-
Continuous Learning: Each fishing trip improves the AI through reinforcement learning
-
Voice-First Design: Complex AI hidden behind simple natural dialogue - "complexity for the machine, simplicity for the human"
Traditional tools cost €20-50K and react to fish presence. Pontos predicts where pelagic schools will be using space-based intelligence, requiring zero hardware investment. While MaxSea provides charts and Global Fishing Watch monitors vessels, only Pontos combines multi-satellite AI orchestration with Deep RL to transform constraints into opportunities - making compliance automatic and sustainability profitable.
Current Status
Completed (Feasibility Study): Surveyed 127 pelagic fishers across France and Costa Rica ports, securing LOIs for 31 vessels worth €48K ARR. Developed thermal front detection algorithm achieving 68% accuracy on historical data. Field-tested PWA prototype with 6 vessels during 3-month trial, validating voice interface (92% recognition in engine rooms). Interviewed 12 cooperative managers, with 4 signing partnership agreements for demonstration phase.
Currently in Progress: Improving AI models to reach an 80% accuracy target through INRIA collaboration using fishing events dataset provided by Direction of Fisheries in Costa Rica. Improving Copernicus’ satellite data processing pipeline. Developing multi-species discrimination for tuna/mackerel/mahi mahi. Building PWA version with offline synchronization.
Starting Q4 2025: ESA Demonstration Project deployment with 100 vessels across France, Costa Rica, and the Spanish coast. Integration with Electronic Monitoring systems onboard. Launch fleet coordination platform for 3 partner cooperatives. Seed fundraising targeting €2M for commercial rollout.
Prime Contractor(s)
