Reference Shelf
A working library of tools, sources, and useful systems thinking
This is the curated side of the site: references worth keeping close when you are planning AI adoption, choosing tools, or improving daily operations without adding clutter.
Curated by hand
Ops-minded links
Useful before impressive
All Resources
AI Fundamentals
Attention Is All You Need
arxiv.orgThe 2017 paper that introduced the transformer architecture underpinning every modern LLM. Understanding why AI works starts here.
Tools & Platforms
LangChain
www.langchain.comFramework for building LLM-powered applications. Most production RAG pipelines start here.
Learning Resources
Elements of AI
www.elementsofai.comFree course from the University of Helsinki making AI accessible to non-technical audiences. Widely used across Nordic organisations.
Norwegian AI Ecosystem
Digdir — Norwegian Digitalisation Agency
www.digdir.noPublishes Norway's national digital strategy and AI guidance for both public and private sector. Context for how government AI shapes market expectations.
Startup Resources
Google for Startups
startup.google.comGoogle Cloud credits and mentoring. Useful if your stack includes Vertex AI or GCP services.
AI Fundamentals
One Useful Thing — Ethan Mollick
www.oneusefulthing.orgPractical AI research communicated for decision-makers. One of the most signal-rich newsletters on AI in real work.
Tools & Platforms
LiteLLM
litellm.aiUnified proxy for 100+ LLM APIs. Simplifies multi-provider AI architecture and gives you usage tracking in one place.
Learning Resources
Prompt Engineering Guide
www.promptingguide.aiComprehensive reference for prompt engineering techniques. Practical for anyone using LLMs day-to-day or building on top of them.
AI Fundamentals
AI Now Institute
ainowinstitute.orgIndependent research on the social implications of AI. Balanced reading for founders thinking about trust and accountability.
AI Fundamentals
Anthropic's Core Views
www.anthropic.comAnthropic's published thinking on safe AI development. Relevant context for any team choosing a foundation model provider.