Year 1 Prototype Priorities
Twenty-one prototypes to demonstrate NORAI capabilities and establish Canada as a global AI leader: including 6 quick-wins and 7 world-leading innovations
Quick Wins (6 Prototypes)
High-impact projects deliverable within 3-6 months
Compute Queue Predictor
ML model predicting wait times across Alliance clusters, suggesting optimal submission times and resources, with auto-retry for failed jobs.
BorealisChat
RAG-based bilingual (EN/FR) chatbot trained on Canadian scientific literature, Tri-Council funding guidelines, and government reports. Hosted on 100% Canadian infrastructure.
Carbon-Aware Scheduler
Route batch jobs to lowest-carbon regions (Quebec hydro vs Alberta gas). Time-shift non-urgent jobs to low-carbon periods. Carbon footprint reporting.
Grant Writing Assistant
AI assistant trained on successful Tri-Council grants (anonymized), understanding NSERC/CIHR/SSHRC formats, reviewer criteria, and common pitfalls.
Research Impact Tracker
Automated tracking of dataset citations, code reuse, downstream discoveries, policy citations, and media mentions.
Collaboration Matcher
Knowledge graph of researchers, publications, datasets, grants with AI-powered similarity matching and collaboration recommendations.
Core Infrastructure (8 Prototypes)
Computome: The Compute Connectome
Unified gateway and intelligent dispatcher for all Digital Research Alliance resources. Automatic routing based on queue times, data locality, and carbon footprint.
Problem: Students wait weeks for campus clusters while national resources sit at 60% utilization. Startups pay $3/GPU-hour on AWS but cannot access Alliance resources.
Outcome: National cluster utilization increases from 60% to 80% ($50M value/year). Startups access HPC at $1/GPU-hour. Students get same priority as professors.
Federated Data Catalog
Unified search API across all Canadian scientific repositories with FAIR metadata and OCAP compliance. Single interface to discover datasets across 13 provincial and federal systems.
Problem: Researchers spend 40-60% of their time searching for datasets across fragmented provincial and federal systems.
Outcome: Data discovery time reduced from weeks to minutes. 100+ datasets searchable in first release.
Consent-as-Code (OCAP Engine)
Machine-readable consent policies with blockchain-based immutable audit logs. Community-controlled dashboards for Indigenous data governance.
Problem: No existing platform automates Indigenous data governance (OCAP) at scale. Manual compliance is slow and error-prone.
Outcome: First platform worldwide to automate OCAP at scale. Competitive advantage over US Genesis initiative.
Provincial Data Sharing Protocol
Automated governance framework with legal compliance built-in for PIPEDA, provincial health acts, and OCAP requirements.
Problem: Cross-provincial research requires 6+ months of legal paperwork and manual compliance checks.
Outcome: Cross-provincial projects go from 6 months bureaucracy to 2-week digital workflow.
Critical Minerals AI
AI-powered geological analysis combining satellite imagery, geophysical surveys, and historical data to identify economically viable rare earth and lithium deposits.
Problem: Canada needs to identify and develop critical mineral deposits to support clean energy transition and reduce dependence on foreign supply chains.
Outcome: 10 new deposit candidates identified. $15B+ potential export revenue. Support for clean energy supply chain.
PermafrostGPT
High-resolution predictive modeling for northern infrastructure resilience. Machine learning models trained on 40+ years of ground temperature data.
Problem: Northern infrastructure faces billions in damage from permafrost degradation. Current monitoring is sparse and reactive.
Outcome: 1-km resolution forecasts. $8-12B projected savings in infrastructure protection across NWT and Yukon.
Privacy-Preserving Health Analytics
Federated learning framework enabling cross-provincial cancer research without centralizing patient data. Secure multi-party computation across provincial data enclaves.
Problem: Provincial health data cannot be centralized due to privacy laws, blocking pan-Canadian research collaboration.
Outcome: Analysis of 10M+ patient records with zero data centralization. 3 novel therapeutic targets identified.
NORAI Data Gateway
Unified API wrapper federating 30+ Canadian open science platforms (CONP, Ocean Networks Canada, CADC, Polar Data Catalogue, CanDIG, FRDR, and more) with native PyTorch/TensorFlow integration for AI/ML pipelines.
Problem: 30+ Canadian open science platforms exist but are siloed. Researchers waste 40-60% of time on data wrangling. No unified way to feed Canadian data into AI/ML pipelines.
Outcome: Single API to discover, access, and stream data from all platforms. Direct DataLoader integration for ML training. Automatic format conversion (DICOM, NetCDF, VCF to Parquet/Zarr).
World-Leading Innovations (7 Prototypes)
2 World Firsts, 5 Global Competitive Advantages
Arctic Digital Twin (Borealis-Earth)
First km-resolution digital twin of Canadian Arctic and boreal ecosystems, integrating permafrost, sea ice, wildlife corridors, and infrastructure. No equivalent exists globally.
Problem: Canada has 40% of the Arctic but no dedicated high-resolution digital twin. NVIDIA Earth-2 and EU DestinE focus on global/European climate.
Outcome: Enables predictive infrastructure planning, supports Arctic sovereignty, powers climate adaptation policy. Positions Canada as Arctic AI leader.
Autonomous Materials Discovery Lab
Self-driving laboratory for critical minerals and clean energy materials. Closed-loop AI + robotics: AI designs experiments, robots execute, ML analyzes, loop repeats 24/7.
Problem: US DOE Genesis Mission is building autonomous labs at Argonne/Berkeley. Canada has no equivalent. Manual materials discovery takes 10-20 years per new material.
Outcome: 10x faster materials discovery. Reduces critical minerals import dependence. Positions Canada in global battery/clean energy race.
Indigenous AI Governance Platform
Complete technology stack for Indigenous data sovereignty: community-controlled consent portals, automated CARE/OCAP compliance, benefit-sharing smart contracts, data repatriation tools.
Problem: No platform anywhere automates Indigenous data governance at scale. CARE Principles and OCAP exist as frameworks but require manual implementation.
Outcome: World's first: exportable to Indigenous communities globally. Positions Canada as ethical AI leader. Addresses reconciliation commitments.
Borealis-Science Foundation Model
Sovereign 70B+ parameter foundation model trained exclusively on Canadian scientific corpus, bilingual (EN/FR), with optional Indigenous language support.
Problem: All scientific AI assistants (GPT-4, Claude, Gemini) are US-based with no understanding of Canadian context, French scientific literature, or Indigenous knowledge.
Outcome: Sovereign AI capability. Serves underserved French research community. Demonstrates 'Made in Canada' AI. Exportable to Francophone nations.
Federated Genomics Network
Privacy-preserving federated learning network connecting all provincial cancer registries and genomic databases: analyze 10M+ patient records without any data leaving provincial jurisdiction.
Problem: Canada's provincial health system creates data silos. Cancer/genomics research requires combining data across provinces, but privacy laws prevent centralization.
Outcome: Enables pan-Canadian precision medicine research impossible today. 3+ new therapeutic targets. Model for global federated health AI.
Borealis-Scholar AI Research Agent
Open-source AI research agent with superhuman literature search and synthesis, trained on Canadian research corpus, fully bilingual, understanding of Tri-Council ecosystem.
Problem: FutureHouse and Elicit (US) dominate AI-powered literature synthesis. No open-source, bilingual alternative exists.
Outcome: Reduces literature review from weeks to hours. Serves French research community. Open-source builds Canadian AI ecosystem.
Quantum-Classical Science Bridge
First national platform seamlessly connecting quantum processors (Xanadu photonic, D-Wave annealing) with Alliance HPC clusters for hybrid scientific workflows.
Problem: Canada has world-leading quantum companies (Xanadu, D-Wave) but no integration with national HPC. No national platform exists to make quantum accessible.
Outcome: Democratizes quantum access. Accelerates quantum advantage timeline. Keeps quantum talent in Canada. First-mover in quantum-HPC integration.
Year 1 Investment Summary
Partner on Prototype Development
We're seeking government partners, research institutions, and industry collaborators to co-develop these foundational capabilities.
Discuss Partnership