This award recognizes an organization that has developed and implemented a groundbreaking AI solution, demonstrating exceptional technical innovation. The winning solution must showcase advanced AI capabilities and illustrate how these technologies have been effectively applied to address complex challenges. This award honors not only technical excellence but also the scalability of the solution and its tangible impact on business outcomes or industry practices, setting a new standard for AI-driven innovation and establishing a benchmark for future technological advancements.

We are technology pioneers who have been digitalizing societies for the past 170 years. Today, our 19,000 talented colleagues serve 25 million customers across the Nordic and Baltic regions with essential digital infrastructure and digital services that are fundamental enablers of the digital societies we live in. We are the telecommunications leader in the region, the leading Nordic media house, and the leader in ICT in both Finland and the Baltics.

Telia started it generative AI initiative almost a year and half ago, where we started with data platforms to accommodate AI workload. During 2024, we are now gaining the fruits of these efforts.

We succeeded in building strong data platforms in both cloud and onprem to help the organization to host and run GenAI applications and even host open source LLMs inside Telia data centers.
We succeeded in implementing multiple GenAI use cases, most important ones are:

– LLMs gateway/library: we built LLMs gateway where using one endpoint, our users can get access to all LLMs we host in our infrastructure (Azure, Bedrock, onprem..etc). This abstraction layer helped the AI developers to easily integrate any LLM in their applications.
– RAG as a service, easier self-service for creating chatbots, i.e. customer support chatbot, confluence documentation chatbot and RFPs chatbots.
– Marketing content generation: empowers content managers and copywriters to produce instant, tailored copy for multiple channels and audience segments
– Metadata generation service: leveraging GenAI to generate column descriptions for datasets. This implementation has been featured during AWS summit and also publicly in AWS customer show case ( https://www.youtube.com/watch?v=wItg1P0zqF0)

The efforts in building such advanced capabilities on both data platforms and applications level throughout 2024 has a huge positive impact in our colleagues day to day work, by improving their productivity using AI and also enhancing our customers experience using our communication channels.

Additional Links
– AWS showcase in youtube: https://www.youtube.com/watch?v=wItg1P0zqF0
– AWS ebook: https://pages.awscloud.com/rs/112-TZM-766/images/AWS_Anthropic_Unlocking_Value_with_Foundation_Models_eBook.pdf
– Presented our onprem and GenAI infrastructure during data summit 2030

The focus in 2023 into 2024 has been to
advertise the Data & AI Strategy as well as to hands-on work with the business to integrate the data goals into their business goals
initiate execution of the strategy on low hanging pain points to build a cache of success stories
showcase a clear red thread of 1 business goal or pain point being resolved through the help of Data & Analytics

The above point 3 is easier said than done since high level business goals are complex and are supported by multiple functions and underlying goals that jointly ensure success. Thus, among the top 4 business goals for Volvo Penta (Business Growth, Customer Success, People and Sustainability) we chose Sustainability as the area to deep dive and focus on.

Sustainability as an area is a green field, with concrete key results and relatively straightforward. Volvo Penta’s top goal within Sustainability is -40% reduction in CO2 emissions. Sudharshan worked together with the other Heads of Data within the Volvo Group to break down the top level Sustainability core business goals into components where Data & Analytics can support.

This has led to extremely positive buy-in from all the Heads of Sustainability within the Volvo Group and focused implementation of various D&A activities (data quality, governance, data engineering). The various areas within Data (Data Governance & Management, Information Architecture, Data Products, Data Literacy, APIs) are also harmonized and working together towards a common business goal. We have seen very good results in holistically working with all areas of Data for a given use case area which also lead to easier adoption by our business as well as manageable workload on the company.

Novo Nordisk is a global healthcare company, founded in 1923 and headquartered just outside Copenhagen, Denmark.
Their purpose is to drive change to defeat serious chronic diseases, built upon their heritage in diabetes. They do so by pioneering scientific breakthroughs, expanding access to their medicines and working to prevent and ultimately cure the diseases they treat. In recent years, AI have been a recent focus area for Novo Nordisk, implementing AI across the organization from drug discovery to production and how best to deliver the products to their patients.
They employ more than 69,000 people in 80 offices around the world, and market their products in 170 countries.

The AI Lab, a team of around 10 individuals, focuses on AI innovation, particularly Generative AI, within the pharmaceutical space. They have developed an internal Generative AI platform and contribute to projects spanning the entire value chain, from research to commercial applications.

The Generative AI platform enhances productivity and innovation, ensuring compliance and interacting with various LLMs. It organizes conversations, saves/shares prompts internally, and integrates/query text documents and tables using code agents. The platform also offers personalized training programs with access to over 40,000 elements, generates images from text, summarizes meetings, and showcases AI capabilities through its Generative AI Playground. These include question-answering, document summarization, content generation, and creating graph networks from documents. With over 30,000 unique monthly users in 2024, the platform documents business value exceeding €40 million annually.

Strategic Projects
Development: The AI Lab has improved access to clinical documents using multi-agent frameworks, streamlining data retrieval and analysis. This enables non-experts to access and analyse clinical data with natural language. This work has been recognized by Microsoft in lightning talks.

R&ED: The AI Lab developed the R&ED Chat, an advanced tool based on the multi-agent framework and connected to several internal/external research databases. It empowers researchers by simplifying information discovery through natural language queries, enhancing data-driven research.

Commercial: We collects extensive medical insights from healthcare professionals (HCPs). The AI Lab created a RAG-based Q&A application allowing frontline professionals to ask questions and receive generated answers from the insights, improving how we interact with with HCPs. Over 90% of users find it highly relevant, with a business value of over €2.5 million per year. The solution is live in over 35 countries.

Established in early 2023, the AI Lab at Novo Nordisk aims to harness (especially) Generative AI technology within the pharmaceutical domain focusing on creating impactful technical solutions.

The primary objectives of the AI Lab are to innovate and integrate Generative AI solutions across Novo Nordisk’s value chain, thereby enhancing efficiency, fostering innovation, and generating substantial business value. The team aims to democratize access to complex data, streamline processes, and empower employees with cutting-edge AI tools.

Key Challenges
Compliance and Data Privacy: Meeting regulatory and privacy standards.
Complex Data Integration: Seamlessly integrating diverse data sources.
Scalability: Developing scalable solutions.
User Adoption: Creating user-friendly AI tools.

Innovative Solution
The AI Lab’s innovative solutions and initiatives have had a profound impact on Novo Nordisk, including

Increased Efficiency: Streamlining processes across clinical development, research, and commercial sectors – and allowing over 30,000 employees to use Generative AI to assists in their day-to-day tasks.
Enhanced Data Accessibility: Empowering employees to access and analyze complex data effortlessly from our clinical trials and internal/external research databases through multi-agent frameworks.
Significant Business Value: Documenting a business value of over €40 million per year from the Generative AI platform alone – with additional value of millions of euro from their specific solutions in R&ED, Development and Commerical.
Global Reach: Implementing solutions that are live, and in production and for the commercial use across, live in more than 35 countries. All in less than 2 years.

The AI Lab’s work exemplifies the transformative potential of Generative AI in the pharmaceutical industry. Their innovative solutions, strategic approach, and substantial positive impact make them a deserving candidate for the Nordic DAIR awards.

The AI Lab has been mentioned in several articles and the work being presented in various conferences around the world in 2024.

Conferences Presentations:
BioTechX Europe, 2024, Basel – https://www.terrapinn.com/conference/biotechx/
WeAreDevelopers World Congress, 2024, Berlin – https://www.wearedevelopers.com/world-congress
Data Innovation Summit, 2024, Stockholm – https://datainnovationsummit.com/speakers/
NDSML Summit, 2024, Stockholm – https://ndsmlsummit.com/

Articles:
Newspaper Finans [Danish] – https://finans.dk/tech/ECE16274747/nu-hegner-novo-nordisk-den-kunstige-intelligens-inde/
Frontcover Story on ComputerWorld [Danish] – https://www.computerworld.dk/art/286599/han-har-udviklet-novo-nordisks-nye-ai-som-allerede-nu-anvendes-af-17-000-ansatte-saadan-har-han-gjort

Mentioned by Microsoft:
https://www.microsoft.com/en-us/research/articles/whats-new-in-autogen/
AI Lab Highlighted by Chi Want in Microsoft Lightning Talk

Podcasts:
Dataklubben, https://podtail.com/de/podcast/dataklubben/s10-ep77-novo-nordisk-chatgpt-blev-et-point-of-no-/
Evo DK: https://evolutionjobs.com/exchange/evo-dk-76-machine-learning-how-to-get-started/

SvenAI, developed by Tyréns AB, is an advanced AI-driven knowledge management platform tailored to meet the rigorous demands of infrastructure and urban planning. This proprietary solution is built to ensure data privacy, enable secure data retrieval, and provide contextually rich insights. Unlike standard AI models, SvenAI integrates Retrieval-Augmented Generation (RAG), Hypothetical Document Embeddings (HyDE), LLM ReRank, and a unique Graph RAG inspired by Microsoft’s “global to local” research. These technologies work together to deliver both broad and detailed responses by navigating relationships within a structured knowledge graph. Key features such as Drag, Drop, Retrain, Collections, Workspaces, and Function Calling empower Tyréns’ team to customize the knowledge base and ensure that SvenAI continuously evolves with user input. This platform has become indispensable for efficient decision-making across complex projects, setting a new standard for secure, AI-driven knowledge management.

Scope of the Project
The SvenAI project aimed to create a secure, adaptable, and precise knowledge management system tailored specifically for the infrastructure and urban planning sectors. Tyréns, an industry leader in these areas, required a solution that could handle complex, context-heavy queries, maintain strict data privacy, and foster collaboration. Recognizing the limitations and risks associated with third-party AI services, Tyréns chose to develop an in-house system that could safeguard proprietary data while leveraging the latest advancements in AI.
Solution
SvenAI combines cutting-edge AI methodologies with user-centric features, creating a powerful, flexible platform:
Advanced AI Techniques: SvenAI incorporates Retrieval-Augmented Generation (RAG), HyDE, and LLM ReRank to deliver accurate, context-aware responses. Graph RAG, based on Microsoft’s “global to local” research, allows SvenAI to structure data into a knowledge graph and navigate complex relationships, providing nuanced answers that typical AI models cannot achieve.
User-Driven Customization: Features like Drag, Drop, Retrain allow users to upload documents and retrain the model to incorporate new knowledge. Customizable Collections and Workspaces let users create project-specific knowledge sets, while Function Calling and Agents enable SvenAI to dynamically respond to diverse query types.
Collaborative Knowledge Hub: File Upload and Commit to Knowledge Structure empower all team members to contribute to the knowledge base, ensuring it stays relevant and comprehensive.
Positive Impact
SvenAI has transformed knowledge management at Tyréns by providing a secure, efficient platform for information retrieval and collaboration. The system’s dual-layered Graph RAG, advanced retrieval techniques, and user-driven customization have streamlined workflows and enhanced decision-making in critical infrastructure and urban planning projects. SvenAI’s modular and adaptable architecture not only reduces reliance on external providers but also adds value by continuously evolving with user input, ensuring it remains a valuable asset across the organization.

SvenAI deserves recognition for its groundbreaking approach to AI-driven knowledge management in the infrastructure and urban planning sectors—a field known for its conservative, risk-averse approach to technology adoption. Infrastructure projects typically involve high stakes, where decisions have long-term financial, social, and environmental impacts. Despite operating in a sector that traditionally avoids innovation due to risk aversion, Tyréns not only built SvenAI to meet stringent internal requirements but also created a solution that outperforms standard offerings across multiple domains.
What makes SvenAI truly remarkable is that it doesn’t just serve Tyréns’ needs. Tyréns has taken the unprecedented step of offering SvenAI as a Software-as-a-Service (SaaS) solution, making this sophisticated, privacy-focused knowledge management system available to other players in the industry, including competitors. By doing so, Tyréns demonstrates a commitment to advancing the entire sector, setting a new standard for AI in infrastructure and urban planning. This move towards openness and collaboration is a bold statement in a traditionally closed industry, showcasing SvenAI’s potential to transform knowledge management across the field.
SvenAI’s innovative approach to balancing data privacy with advanced AI capabilities, along with Tyréns’ willingness to share the platform with others, positions it as a unique solution in a high-stakes industry. It has not only elevated Tyréns’ competitive edge but also established a new benchmark for responsible AI use in sensitive, high-impact sectors. This combination of sector-leading technology, willingness to share expertise, and commitment to advancing industry standards make SvenAI a worthy recipient of this award.

Project Background
Tyréns AB, a leader in the architecture, engineering, and infrastructure sectors, initiated the SvenAI project with support from Predli to address critical data privacy and operational needs. The organization recognized the limitations of mainstream AI platforms, which posed potential risks to sensitive data and lacked the precision required for the intricate, context-heavy queries in infrastructure and urban planning. The goal was to create a proprietary, in-house AI solution that could deliver secure, accurate insights tailored to the specific demands of these fields.
Overview
SvenAI is a comprehensive knowledge management system that integrates multiple advanced AI methodologies, including Retrieval-Augmented Generation (RAG), Hypothetical Document Embeddings (HyDE), LLM ReRank, and a Graph RAG inspired by Microsoft’s “global to local” research. This evolved Graph RAG allows SvenAI to structure data into a knowledge graph, enabling it to navigate complex relationships and provide contextually relevant answers. Key features like Drag, Drop, Retrain, Collections, Workspaces, and Function Calling empower users to continuously expand and refine the knowledge base, ensuring SvenAI’s outputs remain highly relevant and accurate.
Strategic Objectives
The primary objectives of SvenAI were:
Data Privacy and Control: Protect proprietary data and reduce dependency on external AI services, ensuring all sensitive information remains within Tyréns’ secure infrastructure.
Efficient, Data-Driven Decision-Making: Enable quick, precise responses to complex, context-sensitive queries, supporting decision-making in high-stakes infrastructure projects.
User Empowerment and Collaboration: Foster a collaborative environment where users can contribute to and customize the knowledge base, creating project-specific resources for enhanced relevance and accuracy.
Key Challenges
Data Privacy and Security: Relying on third-party AI platforms posed potential risks of data leakage, compromising Tyréns’ competitive edge in the industry.
Handling Complex Queries: Infrastructure and urban planning projects involve highly specialized, context-sensitive questions, which standard AI retrieval models are not equipped to handle effectively.
Innovative Solution
Graph RAG’s Dual Search Strategy: Based on Microsoft’s “global to local” research, Graph RAG combines global and local search strategies within a structured knowledge graph. This approach allows SvenAI to provide comprehensive, contextually rich responses, ideal for the complex, layered queries encountered in infrastructure projects.
HyDE and LLM ReRank: These techniques generate hypothetical documents and re-rank results based on query relevance, ensuring that SvenAI delivers accurate answers even when direct information is sparse.
User-Centric Enhancements: Drag, Drop, Retrain enables users to upload and incorporate new data; Collections and Workspaces allow for project-specific customization; Function Calling and Agents adapt responses dynamically for diverse query types, enhancing SvenAI’s flexibility.
Positive Impact
SvenAI has revolutionized Tyréns’ knowledge management, providing a secure, user-driven platform that simplifies information retrieval and enhances collaboration. By allowing users to contribute to and refine the knowledge base, SvenAI continuously adapts to new information and project-specific needs, supporting better decision-making across high-stakes infrastructure and urban planning initiatives. The platform’s unique Graph RAG approach further ensures that SvenAI can handle complex, multi-faceted queries with precision, giving Tyréns a significant edge in a competitive, risk-averse industry.
Value Added Output in 2024
Throughout 2024, SvenAI has proven invaluable by delivering measurable benefits across Tyréns’ projects:
Enhanced Efficiency: By reducing reliance on external AI providers, SvenAI has streamlined knowledge management workflows, saving time and resources.
Improved Decision-Making: With Graph RAG and other retrieval techniques, SvenAI provides contextually rich answers that directly contribute to higher-quality decision-making.
Collaborative Growth: The platform’s user-driven features have fostered a collaborative knowledge-sharing culture within Tyréns, while its SaaS model has allowed external organizations to benefit from SvenAI’s capabilities, promoting growth and innovation across the sector.
SvenAI’s contributions in 2024 highlight its status as a transformative tool in knowledge management, setting a new standard for secure, user-centric, AI-driven solutions in infrastructure

Building Our Own Knowledge System: Why We Took This Path
https://www.linkedin.com/pulse/building-our-own-knowledge-system-why-we-took-path-stefan-wendin-hxaif/