AI initiative registry
Every initiative listed in the registry can be consulted here, without an account. Each record shows its progress, its milestones and the history of its decisions. Requests still under triage do not appear: they join the registry once reviewed.
Filter
198 current initiative(s)
-
6
ITSD internal chatbotITSD is currently implementing a simple chatbot leveraging Copilot, to streamline access to its knowledge base for our agents. This initiative aims to further improve the speed and quality of service provided by our technicians by enabling them to quickly locate the information they need to provide support and service to clients.
Development Chief Digital Officer Interaction and Engagement
-
191
Irish Moss ResearchAutomatically and non-invasively detect and assess health of Irish Moss Solution being developed for Science to review hundreds of hours of video for Irish Moss presence and health.
Ideation On hold Indigenous Affairs, Aquaculture and Governance Perception and Understanding
-
114
InsightBridge: Executive CSAS SummariesUses AI to convert complex, technical CSAS reports into concise, plain-language summaries tailored for executive decision-makers. Enhances leadership understanding and accelerates informed decision-making by providing easily digestible insights. Ensures executives can quickly grasp critical information without sifting through dense documentation.
Pilot Ecosystems and Oceans Science Cognitive and Generative AI
-
113
Improving forecasting for high-priority portsResearch into feasibility of improving accuracy and timeliness of short-term forecasts for coastal regions by combining traditional and AI modelling methods.
Pilot Ecosystems and Oceans Science Automation and Decision Support
-
103
Improved understanding of the processes impacting harmful algal bloomsUsing machine learning algorithms to analyze phytoplankton images from the Imaging Flow Cytobot (IFCB) to recognize and quantify species associated with harmful algal blooms.
Pilot Ecosystems and Oceans Science Perception and Understanding
-
99
Improved freshwater runoff modelling in Eastern Canada to drive ocean modelsUse Neural Networks to post-process streamflow simulations from the WRF-Hydro model, improving their agreement with observational data. These neural networks are applied as a downstream calibration step. Builds synergy between traditional calibration methods and modern AI techniques in hydrologic modeling.
Pilot Ecosystems and Oceans Science Automation and Decision Support
-
20
IM ChatbotChatbot to answer Information Management related questions, based on information available on the IM intranet site. This is a work in progress.
Development Chief Digital Officer Interaction and Engagement
-
25
Help DFO employees navigate CDOS structure and servicesCDOS is a large and complex organisation. Leverage AI to help DFO user access services, understand organization structure, identify SME. Expose know procedures and documentation through natural language. Using enterprise search, agents and other tools and technologies. Integrate existing tools such as Viva topics that already uses AI to identify expert based on multiple communication channels Right now multiple initiatives goes in this direction, but no unified vision, (pocket of knowledge) Currently not a great user experience, cdos internal staff oriented
Ideation Chief Digital Officer Interaction and Engagement
-
192
Harbour AssessmentsAutomatically and continuously assess harbour conditions via drone and satellite imagery Solution being developed for Small Craft harbours to assess harbour conditions (presence of ice, usage, vessel and vehicle traffic, change detection) based on drone footage and satellite imagery.
Ideation On hold Indigenous Affairs, Aquaculture and Governance Perception and Understanding
-
98
Guidance on Optimal Timing for Environmental DNA (GOTeDNA)Provide guidance on optimal eDNA sampling periods and standardized sampling procedures for assessing and monitoring coastal species using eDNA. GOTeDNA, a centralized interactive online tool currently in development, will report/visualize trends in spatio-temporal eDNA distributions.
Pilot Ecosystems and Oceans Science Automation and Decision Support
-
227
Github CopilotGitHub Copilot is being piloted at DFO aimed at enhancing developer productivity and exploring the operational value of AI-assisted coding. Stats: (90-day period) - 27,584 lines of code were accepted - 17,801 prompts were accepted - Potential savings of $144K ($50/hour developer) based on 40 developers
Pilot Chief Digital Officer Infrastructure and Enablement
-
246
GeoAIOverview: Potential GeoAI‑Impacted Programs and Processes GeoAI (the integration of geospatial data, AI, and machine learning) is being explored to enhance decision‑making, efficiency, and service delivery by automating spatial analysis, identifying patterns at scale, and enabling predictive insights. 1. Program and Service Delivery GeoAI can support more targeted, timely, and evidence‑based services by: Identifying spatial patterns in environmental, socio‑economic, or ecological data Prioritizing areas for intervention, monitoring, or investment Improving accessibility and equity of services through location‑based insights Examples Risk‑based prioritization of inspections, monitoring sites, or enforcement activities Predictive habitat or species distribution mapping to support conservation or fisheries management Enhanced situational awareness for emergency response or environmental incidents 2. Operational Efficiency GeoAI enables automation and scalability in routine or data‑intensive operations, reducing manual effort and turnaround times. Examples Automated feature extraction from satellite imagery, aerial photos, or LiDAR (e.g., shoreline change, infrastructure, habitat types) Near‑real‑time monitoring of environmental conditions or operational assets Change detection to flag anomalies or emerging risks Operational Benefits Reduced reliance on manual digitization and visual interpretation Faster updates to spatial products and dashboards More consistent and repeatable analytical outputs 3. Business Processes and Decision Support GeoAI can modernize internal business processes by embedding spatial intelligence into planning, reporting, and governance workflows. Examples Predictive analytics to support long‑term planning and scenario analysis Decision‑support tools that integrate geospatial, operational, and administrative data AI‑assisted data quality assessment and metadata generation Process Impacts Improved transparency and defensibility of decisions Better integration of spatial data across programs and systems Enhanced performance measurement and outcome tracking 4. Cross‑Cutting Opportunities Across all domains, GeoAI supports: Data integration: Linking spatial, temporal, and non‑spatial datasets Scalability: Applying consistent analysis across large geographic areas Innovation: Enabling new analytical questions that were previously impractical
Ideation Chief Digital Officer Automation and Decision Support
-
26
Generative AI in intranet Internet content creationIntegrate Generative AI capacity right into Drupal CMS to help users create draft content, ease translation, generate images, tags and metadata. Technologie already exists, need to be security reviewed, integrated and configured. Challenge, web infrastructure team capacity
Ideation Chief Digital Officer Cognitive and Generative AI
-
183
General - Office ManagementDraft and Refine Documents: Staff make use of Copilot to generate or refine briefing notes to make these clear and concise. Enhance Collaboration and Workflow Efficiency: By offering suggestions for document formatting, language style, and clarifying content, Copilot helps maintain consistency and accuracy across communications and facilitates more effective interdepartmental collaboration. Meeting notes: After a meeting is recorded, Copilot can transcribe audio to produce a relatively accurate record of the discussion, using natural language processing to identify key points, decisions, and follow-up tasks, and organizing these into relatively concise summaries and action item lists. Conduct Research and Summarize Information: It can quickly extract key points from extensive reports, legislation, or guidelines—helping to synthesize complex information into actionable insights. Support Data Analysis and Reporting: Employees use Copilot to structure performance metrics, track progress against key indicators, and generate reports that inform decision-making. **These functionalities save time, improve comprehensiveness, and can help ensure that critical details are captured.**
Pilot Programs Sector Cognitive and Generative AI
-
208
Fishing DetectionThe goal is to use vessel movement behaviour in the form of AIS data to predict whether a vessel is currently engaged in fishing activity. AIS data is available to the department in near real-time from AIS transponders equipped on vessels. An automated system that can indicate to fishery officers when and where vessels are likely engaged in fishing activities will support better monitoring of compliance with regulations such as conformance with conditions in fishing licenses and in marine protected areas. Due to limited capacity and lengthy processes to manually review and distill data sources, fishery officers cannot fully monitor all vessels under present circumstances. This AI-supported system would help to increase monitoring coverage for fishery officers. Spatial processing: The vast amounts of geographically-referenced AIS transmissions from vessels must be refined to spatiotemporal areas of interest prior to feature engineering for input to the machine learning model. Appropriate spatial processing can support achieving this in a timely manner, particularly when near real-time processing is of interest. However, given the availability of the AIS pipeline, this processing is more appropriate to be performed by the pipeline than within the analytics environment for this use case. The fishing detection system would use the AIS pipeline API and would received data which has already been refine to the relevant spatial extents.
Pilot Strategic Policy Automation and Decision Support
-
219
Fish-Habitat AssociationsExtraction of relevant habitat association info and strengths based on vetted literature input to AI for synthesis. Goal to improve equivalency models used in regulatory decisions. Possible partnership with McMaster University.
Ideation Ecosystems and Oceans Science Cognitive and Generative AI
-
123
Fish-Habitat AssociationsExtraction of relevant habitat association info and strengths based on vetted literature input to AI for synthesis. Goal to improve equivalency models used in regulatory decisions. Possible partnership with McMaster University.
Pilot Ecosystems and Oceans Science Cognitive and Generative AI
-
106
Finwave: an online photo-identification database and AI matching toolThis project aims to improve a publicly accessible machine learning platform capable of identifying individual killer whales in real-time from submitted photographs. The ultimate goal is to improve the web-based interface and validate the algorithm used for photo identification against known individuals.
Pilot Ecosystems and Oceans Science Perception and Understanding
-
257
Figma AIThe initiative explores the use of Figma's AI capabilities to support user experience (UX) design activities across DFO. Figma AI features may assist in generating interface designs, creating wireframes, producing placeholder content, organizing design assets, accelerating prototyping activities, and improving collaboration among design teams. The objective is to improve productivity and reduce manual effort while maintaining human oversight of all design decisions and outputs.
Ideation Chief Digital Officer
-
144
Factoid Finder ToolThis completed project used the Factoid Finder extracts text from machine-readable PDFs to create searchable content libraries. Small Language Models are used to implement semantic search techniques, allowing users to search the library for relevant passages within the PDFs.
Pilot Ecosystems and Oceans Science Cognitive and Generative AI
-
63
Factoid Finder ToolThe Factoid Finder extracts text from machine-readable PDFs to create searchable content libraries. Small Language Models are used to implement semantic search techniques, allowing users to search the library for relevant passages within the PDFs. Small Language Model and Text Embedder: MS Marco Distilbert Dot-v5 Cross-encoder: MS Marco MiniLM-L6-v2
Pilot Cognitive and Generative AI
-
72
Extracting impacts data from regulatory documentsUse OCR pdf readers, topic detection and language models to detect specific data elements inside written text documents, for extraction and analysis in a cumulative impacts context. Large lanugage models and topic clustering tools
Ideation Ecosystems and Oceans Science Perception and Understanding
-
258
Establishing a Standardized National Biotelemetry Data PipelineDevelopment of a national biotelemetry platform that standardizes the processing of acoustic animal tracking data. Automated workflows that apply Random Forest and neural network methods to identify movement patterns, mortality events, and other biologically significant behaviours will standardize quality control, improve data interoperability, and provide self-service analytical tools for scientists and managers across multiple departments, supporting faster evidence-based decision making.
Development Ecosystems and Oceans Science
-
51
EN/FR Translation ModelA large language model that is trained on the Translations Bureau’s and DFO's language-characteristics (i.e. grammar, annotation, syntax, words, scientific term) context. The model’s enables translation to be like those by the Translation Bureau. Training on large database of scientific documents (published by Canadian Science Advisory Secretariat) to understand scientific terminology is currently in progress. MADLAD 400 10B (Initially trained on sample of ~100 translated documents, in progress of additional training with published CSAS documents going back to year 2000)
Pilot Programs Sector Interaction and Engagement
-
77
Emulator for OPP port models and for relocatable ocean modelling systemUse machine learning to develop an emulator (or similar) for the OPP port models, which would be used to produce forecasts via inference. The inference is expected to be both faster and computationally cheaper than existing solutions. The training process would use existing model output from multi-year hindcasts. Success with a fixed location would graduate to developing a relocatable emulator TBD; could be U-Net or CNN, or based on GraphCast
Ideation Ecosystems and Oceans Science Cognitive and Generative AI