Standards Norway

Case study
Bringing AI to Norway's standardisation landscape

Improving processes with AI

Standards Norway partnered with Reodor Studios to bridge the gap between AI potential and real-world usability—ensuring that AI didn’t just exist as a proof of concept but created measurable improvements in efficiency. Unlike typical AI experiments that stay in the pilot phase, Reodor’s ‘prove, then improve’ approach ensures that solutions don’t just work in a test environment but are scalable and adaptable to real industry needs.
Standards are essential, but complex. Users struggled daily with finding, understanding, and applying the right standards to their work. The traditional approach of manually searching through PDFs is inefficient as they have to decode complex language, connect dots across regulations, and distill dense documents into actionable insights. Standards Norway needed a reliable and accurate solution that could make their vast knowledge base more accessible and user-friendly. Our proof of concept approach validates feasibility, mitigates risk, and shapes the MVP for testing – making Reodor the right partner for this project.
  1. Accelerating the green transition for industrial customers
  2. Creating digital, scalable business models to complement existing consulting services
  3. Building trust in emerging technology
  4. Identifying strategic opportunities for new service development

The challenge

Before building a solution, we had to ask: what were the real challenges users face when using technical standards? We started with conducting user interviews, observing workflows, and analyzing pain points – just real people sharing real challenges. A clear pattern emerged:
  1. Users struggled to determine whether a standard was relevant to them.
  2. Finding the right standard was time-consuming.
  3. Once they found it, they had to locate a specific section, interpret complex language, and connect it to other relevant regulations.
These challenges aligned well with AI’s ability to retrieve and contextualize information. The insights guided us in the right direction – from simple no-code prototypes to a sophisticated and precise AI-powered concept.

The approach

Our development followed a normal "prove, then improve" strategy:
  • Started with a simple no-code prototype in Appfarm connected to OpenAI's Assistant API
  • Validated core concepts through several user tests and feedback sessions - both internal and external
  • Built a basic user interface to track conversations and user patterns
  • Scaled up with more standards content and refined the model
  • Finally, implemented advanced RAG architecture (retrieval augmented generation) for precision and reliability
The results? For Standards Norway, this isn’t just about AI—it’s about modernizing how professionals access and apply technical standards, reducing time spent on compliance and unlocking new opportunities for industry-wide digital transformation.
Tech stack
  • Frontend: Next.js with Tailwind CSS for a responsive, modern interface
  • Backend: Supabase with pg_vector for efficient vector storage
  • AI integration: OpenAI's embedding models and Assistant API
  • RAG implementation: Custom vector database for precise information retrieval
  • Data processing: Advanced chunking and embedding pipeline for standards content
“Our proven Proof of Concept product, which validates feasibility, reduces risk, and defines the MVP, was a perfect match in this case. Since the challenge of quickly finding and interpreting information in standards is a common pain point across industries, we had an insight-backed problem to solve—allowing us to jump on it and execute.”
Magnus Bjelland, Creative Technologist, Reodor Studios

The results

Transitioning from manual searches in PDFs to an AI-powered assistant is a major step forward, and we’ve built a solution that delivers precise, context-aware results. We provided a recommendation on how it could be implemented, delivered the data and all AI-generated results, allowing them to shape the path forward themselves. They have structured the recommendation into two stages and are currently implementing what they refer to as "Eivind light" in-house, while looking forward to partnering with us on the commercialization of the service.
While we're excited about the current capabilities, we're even more thrilled about the future possibilities this opens up for Standards Norway and its users, as part of the larger transformation journey they are on.