The Fintraffic Mobile app provides real-time information on Finnish roads and rail traffic. Information on traffic disruptions and warnings are sent as push notifications directly to your phone – wherever you are. You can also use the application to report any traffic problems and disruptions you encounter.
How does the app work?
Fintraffic App uses real-time data obtained directly from Fintraffic’s traffic centres, so you always get up-to-date information on the traffic situation. By allowing push notifications from the app, you can receive traffic announcements based on your location even while you are travelling.
The convenient map view shows you what is happening on the roads in your local area. You can choose which items you want to see on the map. The application includes all Finnish road weather cameras, road works, congestion points, driving conditions and weather information, road weather warnings, electric car charging stations and gas (CNG) car refuelling stations and a host of other useful information.
The rail traffic section offers a fully real-time view of all long-distance and regional traffic timetables and live train updates. For example, you can save your business trip as a favourite route and you will be informed immediately if there are any disruptions or changes to the route.
You can easily submit your feedback to the Traffic Customer Service’s Feedback Channel service using the application. Have you seen a pothole in the road or a damaged traffic sign? Is one of the announcement screens on a train station platform out of order? The Traffic Customer Service processes feedback and forwards the information to others, including contractors that carry out maintenance work.
You can report traffic deviations and hazards, such as animals wandering along the road, stopped vehicles, poor driving conditions and other unusual traffic conditions. User reports will appear on the map for 30 minutes. Reporting is easy.
YODA is PostNord’s unified, governed data and AI platform designed to consolidate fragmented Nordic data landscapes into a single scalable foundation for BI, ML, and AI use cases. The initiative was launched to address structural inefficiencies across 13 separate data warehouse environments, inconsistent KPI definitions, duplicated infrastructure costs, and slow time-to-market for cross-domain analytics.
The scope of YODA includes:
Establishing a centralized, cloud-native lakehouse architecture on Azure and Databricks
Harmonizing data definitions across countries and domains
Enabling secure self-service analytics for business domains
Providing a governed foundation for ML and AI initiatives
Implementing standardized ingestion, transformation, and semantic modeling patterns
Introducing enterprise-grade data governance and access control
YODA operates under a federated data mesh-inspired model, where domains own their data products while leveraging shared platform capabilities, security guardrails, and standardized golden paths. The platform supports batch and streaming ingestion, curated medallion architecture (bronze, silver, gold), and enables both traditional BI workloads and advanced AI/ML pipelines.
In 2025, YODA matured from a consolidation initiative into a strategic AI enabler. Key developments included:
Expansion of cross-domain data products for logistics optimization
Enablement of ML use cases such as parcel delay prediction and route optimization
Standardized governance mechanisms aligned with AI risk classification and EU regulatory considerations
FinOps-driven cost transparency and optimization across compute workloads
Increased self-service onboarding of domain teams through structured enablement models
The positive impact includes:
Reduction of infrastructure and licensing redundancy
Significant improvement in cross-country KPI alignment
Accelerated time-to-market for analytics initiatives
Foundation for scalable AI adoption across
In 2025, Novo Nordisk accelerated an enterprise-wide AI transformation that spans research, development, and enterprise operations.
The company’s AI platforms unify data and models, provides a copilot for advanced reasoning, enables governance of “reasoning chains,” and functions as an “AI factory” to build, reuse, and fine-tune models across use cases spanning regulatory affairs, early research, discovery, and trial design. Early outcomes include predictive AI models for advanced cardiovascular risk detection that outperform leading clinical standards, pointing to meaningful clinical impact for precision medicine. [1] Another key result is a Co-Scientist ecosystem — an agentic AI layer that surfaces clinical insights in seconds, democratizing access to data (now used by over 20% of development), automating workflows, and improving trial design. It is expected to deliver >$157M in net new value and >$14M in operational efficiency gains over the next five years. [2]
To meet compute demands, Novo Nordisk leverages the Gefion national AI supercomputer (funded by the Novo Nordisk Foundation) to scale training and simulation for drug discovery—enabling larger experiments in protein engineering, biological modeling, and generative molecule design. In parallel, a strategic collaboration with NVIDIA advances customized GenAI models and agentic workflows for early research and clinical development, including single-cell response prediction, drug-like molecule generation, and biomedical LLMs powered by NVIDIA BioNeMo for computational drug discovery. This collaboration includes several AI research programs, including using single-cell models to predict cellular responses to drug candidates and structures, as well as designing models to build molecules with drug-like properties. The companies will also collaborate on tapping Novo Nordisk’s vast global scientific literature to build biomedical large language models, enabling researchers to uncover correlations between genes, proteins and diseases. [3]
The company is executing a company-wide Agentic AI strategy. Internally, agentic frameworks such as Nautilus (assisting researchers), FounData (assisting with analysis of clinial data), MedIQ (assisting field workers) and Biological Reasoning (contributing to drug discovery) are already deployed, with targeted initiatives to replace entire specialized research workflows with agentic AI to accelerate the scientific cycle. Looking ahead to public science impact, Novo Nordisk will present models at ECO 2026 that support innovative identification of the highest-need patients for obesity treatment—work with potential to significantly reduce obesity related comorbidities, a significant challenge to society. [4] Responsible AI is a standing priority, overseen by an internal committee that guides safety, governance, and ethics across platform, agents, and use cases [5]
The company also operates a GenAI platform with 30,000+ users, access to leading foundation models from major providers, and more than 250 billion tokens consumed in 2025—on the order of 40 times the full text scale of Wikipedia—demonstrating production-grade adoption and impact at scale. [6, 7]
To expand frontier capabilities, Novo Nordisk is advancing partnerships that explore multi-agent systems and foundation models in real-world data (RWD) contexts. A collaboration with Google Health is evaluating how multi-agent systems can enhance data science productivity and uncover new RWD-informed hypotheses. On the modeling frontier, Novo Nordisk is progressing protein LLMs to predict attributes and functions from molecular sequences, informing target discovery, protein engineering, and developability assessments. In addition, they entered a partnership with QPurpose to explore potential in Quantum Compute in relation to e.g., AI modelling. [8]
Organizationally, Novo Nordisk appointed a Chief AI Officer and executed a re-organization that places Enterprise AI at the center of its operating model—aligning leadership, funding, security, and accountability behind AI-enabled value creation. Together, these moves strengthen governance and scale for agentic systems, LLMs, and scientific AI. [9]
Altogether, Novo Nordisk’s 2025 achievements demonstrate a coherent, outcomes-focused AI operating model: a governed GenAI platform reaching tens of thousands of users; Gefion supercomputing for large-scale training and simulation; NVIDIA collaborations for customized models and agentic workflows; and a responsible AI governance structure. The result is faster, safer discovery cycles; improved reproducibility and explainability; and clinically relevant advances with potential for population-scale benefit in areas like cardiovascular and obesity care.
ABS (Asset Backed Securitization) IT Service is a cross‑functional data, analytics and AI Product designed to move from fragmented, local reporting to a governed, scalable capability that enables faster decisions and measurable business outcomes. The Product focuses on creating a trusted data foundation, industrialized analytics, and repeatable ways of delivering AI‑enabled improvements across TRATON Financial Service, while meeting strict requirements on security, privacy, and regulatory compliance.
Scope and solution
– Business scope: Prioritized end‑to‑end flow within the funding process where data gaps and manual work limited performance (e.g., planning, operational follow‑up, quality, risk/compliance). The scope includes both decision support and operational use (embedding insights into daily routines and systems).
– Data scope: Selected critical data sources (internal systems and approved external data) were mapped into a common, well‑documented model. Automated pipelines improved data quality, lineage, and refresh frequency. Data products were created with clear ownership, SLAs, and access controls.
– Analytics & AI scope: On top of the data products, the team delivered a set of high‑value use cases such as analytical insight, anomaly detection, portfolio optimization and advanced KPI follow‑up. Models and analytical logic are versioned, tested, monitored, and continuously improved using an industrialized analytics/ML lifecycle.
– Operating model: A product‑oriented way of working was introduced (cross‑functional squads, backlog prioritization with business owners, and standard templates for value tracking). Governance covers data ownership, information classification, and AI considerations, with an emphasis on reuse and scalability.
Positive impact
The ABS IT Service has improved the organization’s data readiness and ability to scale analytics and AI. Early results include:
– Faster, more reliable decision‑making through a single source of truth for key metrics
– Removed manual reporting and reconciliation by and improved data quality (fewer defects, clearer ownership)
– Shorter lead time from idea to production for reporting analytics/AI use cases, from months to days
– Measurable business outcomes in prioritized processes: Data & Analytics first changed the business process from re-active to proactive with a proven monetization.
– Increased adoption: Product teams, supported by training and enablement that strengthens long‑term capability
Overall, ABS IT Service demonstrates how a modern data foundation, robust governance, and a pragmatic delivery model can turn data into sustained business impact—and create a scalable platform for future financial AI‑driven innovation
During 2025, Epiroc executed a major transformation in how the company operates, makes decisions, and creates value through AI and Data. Just one year earlier, the company had no enterprise AI capability, no AI governance framework, and no AI solutions in production. Analytics initiatives were fragmented and largely exploratory.
Within twelve months, Epiroc established a scalable enterprise AI & Data capability and embedded it into core business operations.
To enable this shift, the company redesigned how AI and data initiatives are owned and developed across the organization. A federated Hub-and-Spoke operating model was introduced, enabling innovation close to business operations while maintaining enterprise standards for governance, quality, and compliance. This structure allows AI solutions to scale responsibly across a global organization while remaining aligned with operational needs.
As a result, Epiroc rapidly built an AI delivery portfolio. Within the first year, 17 AI initiatives were launched, with four solutions already in production and two scaled globally across business operations. These AI capabilities are embedded directly in service and supply chain processes and are delivering measurable operational value by improving productivity and reducing time spent on manual analysis and troubleshooting.
Two flagship solutions illustrate the transformation:
Technician CoPilot, an AI assistant supporting service technicians globally in diagnosing equipment issues by searching technical documentation and recommending solutions in seconds.
An AI capability embedded in supply chain operations, improving operational decision-making and efficiency.
Both solutions moved from concept to production within months and are now used across multiple regions, demonstrating Epiroc’s ability to industrialize AI delivery and move beyond experimentation.
Another key element of the transformation has been shifting from centralized analytics delivery toward self-service analytics in the business. During 2025 this new way of working was successfully implemented within the Global Sourcing function. By establishing clear ownership of analytics within the function, supported by reusable data models and governance structures, sourcing teams gained faster access to trusted insights while significantly reducing dependency on central BI teams.
At the same time, Epiroc introduced a federated AI governance framework to ensure that scaling AI across the organization is done responsibly. The framework combines centralized oversight for ethical guidelines, risk management, and regulatory alignment with decentralized data ownership in business domains. This enables innovation while maintaining transparency and compliance.
Generative AI has also played a significant role in increasing personal productivity across the workforce. During 2025, more than 1,000 employees completed a six-week AI micro-learning program, and over 2,000 employees actively use Microsoft Copilot in their daily work. In Microsoft’s Copilot utilization index, Epiroc reached a score of 71 compared to the customer average of 54, demonstrating strong adoption of generative AI capabilities.
To support this transformation, Epiroc invested significantly in internal engagement and capability building. A newly launched internal AI & Data community quickly became the largest community in the company, fostering knowledge sharing and collaboration. In December 2025, Epiroc also organized an AI & Data Hackathon in Sweden with over 100 internal and external developers, focused on creating solutions for last-mile delivery optimization.
In a single year, Epiroc moved from having no enterprise AI capability to operating a growing portfolio of production AI solutions, active workforce adoption, and a scalable governance and operating model. This transformation established AI and Data as core business capabilities supporting operational excellence and innovation across the company.
The Riksbank is transforming into a modern, open and resilient central bank that uses data and AI as core capabilities. This development has been driven through a three‑ to four‑year data and analytics transformation, which has now successfully transitioned into the permanent organisation. New ways of working and a fully cross‑functional governance model are firmly established.
A major part of the transformation has been building a modern data and analytics ecosystem that functions reliably under strict regulation. Since the Riksbank cannot freely use public cloud services, the data platform is built on premises using open source components, strong security controls and container based technologies. This ensures flexibility between environments and providers while keeping sensitive information protected and enabling advanced analytics and AI.
To strengthen analytical capacity, the Bank defined four ambitions:
1. Carry out an analytical shift based on more appropriate, high quality and timely data.
2. Use new tools and data to increase the capacity for data driven decision making.
3. Create a new technology infrastructure enabling stability, sovereignty and flexibility.
4. Explore new analytical opportunities using modern technologies, including AI.
A unified environment for data, code and reports enables efficient collaboration, reproducibility and version controlled analytical work. Teams work in a fully integrated value stream that brings together data engineers, platform engineers, data scientists, AI engineers, software developers, IT security specialists and architects with economists, analysts, legal experts and subject matter experts.
The underlying architecture uses open metadata standards for governance and a structured data approach inspired by the medallion model, ensuring traceability, quality and well managed data flows from raw ingestion to analytical and policy ready products.
Seven dedicated data teams focus on core policy areas of the Riksbank’s mission, including Monetary Policy, Financial Stability, Payments and Markets. These teams work side by side with domain experts to produce high quality data products, models and insights that support critical policy functions.
Security is integrated throughout. DevSecOps practices, supply chain security and continuous testing ensure safe development from the start. Sensitive data remains within protected on prem environments, and the platform architecture supports innovation without compromising trust or compliance.
AI readiness under development:
The Riksbank is also building secure, sovereign and long term AI capabilities. While still under development, the Bank is establishing:
• Secure on prem computational capacity to support modern AI models without cloud dependency.
• Isolated and controlled environments for large language models, ensuring confidentiality and regulatory compliance.
• AI supported internal knowledge retrieval, using advanced techniques to help employees access and navigate internal policies and documents.
• Automated tools that improve accessibility, documentation and analytical workflows.
• A systematic plan for building AI competence, including training, knowledge sharing and practical adoption in day to day work.
Together, these developments ensure that the Riksbank is well on its way to becoming AI ready, enabling future use cases even for sensitive data.
In January 2025, DeLaval — one of the world’s leading dairy farming technology companies — embarked on a bold, enterprise-wide data and AI transformation. Internally named D4D (Data for DeLaval) and affectionately dubbed “Data Therapy,” this initiative set out to fundamentally rewire how DeLaval thinks about, accesses, and creates value from data and artificial intelligence.
The scope of the transformation is broad and deep, touching strategy, operating model, culture, people, processes, and technology simultaneously.
STRATEGIC SCOPE
The initiative began not with technology procurement but with diagnosis. The incoming Senior Data & AI Platforms and Products Lead ran structured data strategy sessions across DeLaval’s Digital Solutions (DS) organization — engaging domain cells , platform teams, and senior leadership including the CIO, LT, and General Management. These sessions surfaced a clear picture: data teams had become organizational bottlenecks, data access was slow and manual, and data duplication was eroding trust in data-driven decision-making. The resulting D4D strategy was co-created with stakeholders, grounded in their lived experience, and endorsed at the highest level of the organization.
OPERATING MODEL
A new hub-and-spoke (decentralized) operating model was designed and implemented. The central Data & AI Platform team (the “hub”) now owns shared infrastructure, governance standards, tooling, and data platform services. Business domain cells (the “spokes”) own their domain-specific data products and analytics, operating autonomously within governed, standardized frameworks. This model simultaneously centralizes governance and decentralizes execution — enabling scale without surrendering accountability.
ORGANIZATIONAL SCOPE
Two specialized teams were formally established within a new Data & AI Product Area:
— Data Platform Team: tasked with modernizing DeLaval’s legacy platform toward a cloud-agnostic, zero-copy architecture supporting AWS, Azure, SAP, and on-premise environments, while delivering self-service capabilities and reusable components to domain teams.
— Cognitive Platform Team: tasked with scaling AI and agentic solutions as-a-service across all product areas, delivering repeatable, governed AI capability rather than one-off experiments.
PEOPLE & CULTURE SCOPE
The “Data Therapy” framework addresses the human side of transformation across three dimensions — People, Process, and Product (Tech). People interventions include upskilling and reskilling programs, experience mapping across data personas, a community of data advocates, data owner incentive programs, and DevEx improvements. Process interventions include golden path frameworks for data personas, data mentorship programs for cells, role boundary clarity, analytics handoff processes, and roadmapping. Technology interventions include self-serve platform development, low/no-code insights capabilities for non-technical teams, standardized engineering practices, platform vitals monitoring, governance implementation, and centralized external data sharing infrastructure.
AI & PRODUCT SCOPE
Within the first 12 months, DeLaval moved from zero production AI to a live AI services model, with first-version solutions releasing in predictive maintenance, service support enablement, and generative AI capabilities — all delivered through the Cognitive Platform team’s as-a-service model.
By 2028, the vision is for every DeLaval team to independently access trusted, compliant, high-quality data through an intuitive self-service ecosystem — seamlessly integrated into DeLaval’s processes, culture, and daily decision-making from day one of any new employee’s tenure.
This is not a platform upgrade. It is a transformation of how a global industrial company relates to its own data — and ultimately, to its customers and the farmers it serves.
Skatteetaten has been executing one of the most ambitious and comprehensive data and AI transformations in the Nordic public sector. This achievement is especially significant given public organisations’ traditionally fragmented, hierarchical, and bureaucratic structures. Despite these barriers, the Agency has succeeded in embedding data and AI at the core of its operations, culture, and long-term strategy—setting a new benchmark for public sector modernisation.
Summary: The “Maximus” Transformation
In 2025, Lomax moved from ad-hoc AI experimentation to a structured, governed transformation designed to scale responsibly and create measurable value. The model was scaled across departments and embedded deeper into operational workflows. The approach was built on two synchronized tracks:
Track 1 — Organizational Anchoring (governance, adoption, and responsible use)
Lomax established a clear operating model to ensure AI is deployed with leadership ownership, accountable execution, and broad adoption:
• AI Steering Group (CEO, Head of HR, Head of Strategy, Head of Digital): prioritization, governance, and strategic alignment.
• AI Group (execution and technical expertise): delivery of AI tools and projects, model training, prompt development, standards, enablement, and departmental support.
• AI Ambassadors (cross-department network): local change agents and subject matter contacts, feeding use cases and driving adoption.
To reduce risk and increase trust, Lomax implemented practical guardrails by ensuring centralized management of AI licenses, internal AI guidelines, and a default human-in-the-loop approach for AI outputs.
Mandatory training was rolled out via a training platform: short modules on AI fundamentals and Lomax AI guidelines, followed by a quiz to ensure understanding. As of 2025, ~80% of employees completed the training and all new hires are required to complete it.
Track 2 — Value Creation (Maximus: AI embedded in workflows)
In parallel, Lomax developed and deployed Maximus: a centralized suite of in-house developed AI tools integrated into day-to-day work. Maximus is depicted as a red Viking-android persona that makes it easier for employees to refer and relate to. This has been a central part of the successful AI adoption across departments. Maximus is designed as modular applications so model components can be upgraded over time, while workflows remain stable.
Key Maximus solutions and outcomes:
• Maximus Product Translator (from Danish to Swedish)
Reduced translation cycle time significantly. Where translating 250 products previously required 4 weeks, the new workflow demonstrated translation of 250 products in 1 hour, lowering time-to-market and reducing dependency on external agencies while freeing capacity for quality control and review.
• Maximus Data Lab (product text optimization at scale)
Automatically optimizes and generates product titles, punchlines, highlights, and descriptions for both Google and the website across 41,000+ products, improving consistency, discoverability, and customer experience.
• Maximus MaxInsights (voice-of-customer + experimentation engine)
Aggregates and analyzes customer feedback across sources (NPS feedback, Trustpilot, on-site surveys) and drives structured experimentation. Using AB Tasty and best-practice testing rules, Lomax executed 30 A/B split tests based on the findings in the customer feedback and the accumulated impact contributed to a 20% site conversion rate uplift (defined as the accumulated CVR effect of implemented winning tests plus avoided potential CVR losses from losing tests that were not rolled out).
• Maximus Media Library (visual quality at scale)
Upscales images and generates lifestyle backgrounds to improve visual consistency and illustrate use cases for a better customer experience.
Architecture pattern across Maximus applications:
Input → Enrichment → Generation → Human Approval → Publishing
Approval is either full review or controlled sampling depending on risk and use case. In the Media Library, every image is approved in a side-by-side module (original vs AI-generated). In Data Lab, controlled sampling is made before approving all text outputs.
Model approach (selected examples):
• Text generation uses OpenAI fine-tuned models (GPT-4o fine-tuning) with continuous iteration based on expert-produced data (e.g., SEO team outputs for Google-optimized product text).
• Customer feedback categorization and translation are implemented modularly so the underlying model can be updated to the newest version.
• Image generation/upscaling uses a Gemini-based image model, applied within an approval workflow.
Adoption and productivity impact
By October 2025, Lomax achieved broad behavioral adoption: 71% of employees used AI tools weekly (defined as using any AI tool weekly—Maximus, ChatGPT, Claude, etc.). Average time savings reached 2 hours per employee per week, equating to 296 hours saved weekly across the organization. Time savings were measured through a survey sent to all office employees (N=176; respondents=132) using a professional survey tool that prevents duplicate responses.
Timeline overview (2023–2026)
2023: Data foundations and ownership established
2024: Secure generative AI introduced and scaled
2025: AI literacy embedded across the organisation
2026: Enterprise AI platforms and operating models in place
Over the past 2.5 years, Íslandsbanki has delivered a coherent Data and AI transformation, moving from foundational capabilities to enterprise‑grade AI adoption—while operating as a small bank in a heavily regulated environment.
Foundations and enablement (2023–2024)
The transformation started with clarifying data ownership, data quality accountability, and decision rights, supported by bank‑wide data and AI policies. These foundations ensured that all subsequent analytics and AI initiatives were controlled, auditable, and aligned with regulatory expectations—without slowing delivery.
Modern data and analytics capabilities
Íslandsbanki strengthened its core data platform, introduced standardised analytical and semantic models, and implemented shared design and quality standards for reporting and dashboards.
These components reduced duplication, improved trust in management information, and enabled consistent insights across business units, supporting both commercial decision‑making and regulatory reporting.
Secure generative AI adoption (2024–2025)
Following regulatory approval, the bank introduced Microsoft Copilot through a phased rollout. Key components included:
Data loss prevention and information classification
Clearly defined usage boundaries
Monitoring and access controls
Copilot was embedded directly into daily work such as document creation, meeting summaries, and internal knowledge retrieval, delivering immediate productivity gains within controlled limits.
AI literacy as infrastructure (2025)
To make adoption sustainable, Íslandsbanki designed and launched a tailored AI Education Program for its ~750‑employee organisation.
The program established a shared baseline for responsible AI use, addressed ethical and regulatory considerations, and supported multiple AI tools under a single governance model. AI literacy was treated as a permanent organisational capability, not a one‑time rollout.
From use cases to platforms (2025–2026)
The bank progressed from individual AI use cases to building repeatable, enterprise‑level AI components, including:
A governed low‑code and automation framework for productivity and administrative processes
Structured support for generative AI agents, with modular design, audit trails, and human‑in‑the‑loop controls
A clear operating model separating low‑risk experimentation from centrally reviewed, higher‑risk solutions
These components allow innovation to scale safely and predictably.
Delivering under constraint
All progress was achieved through incremental delivery, disciplined investment, and reuse of existing platforms. Security, privacy, and regulatory compliance were maintained throughout, without compromising pace.
Together, these milestones demonstrate how Íslandsbanki has responsibly built modern Data and AI capabilities—showing that even a smaller bank can achieve meaningful, organisation‑wide AI transformation