Insights

The sources we rely on

Stanford HAI — AI Index Report

What it is: The most authoritative annual stocktake of AI, now in its ninth edition — 400+ pages of vetted, globally sourced data on capability, investment, adoption, and policy.

Why it matters for you: This is the reference your board will recognise. We use it to separate genuine market shifts from hype, and to benchmark where your sector actually stands.
→ hai.stanford.edu/ai-index

arXiv (cs.AI / cs.LG / cs.CL)

What it is: The open repository where almost all frontier AI research appears first, often months ahead of formal publication.
Why it matters for you: It tells us what's coming before it reaches your competitors' roadmaps. We monitor it so you don't have to.
→ arxiv.org

Hugging Face — Papers & Hub

What it is: The open commons for AI models, datasets, and paper discovery — and the practical successor to Papers with Code, which closed in 2025.
Why it matters for you: It's where we assess what's genuinely usable today versus what's still a lab result. Relevant when the question is "can we build this now?"
→ huggingface.co

Stanford HELM

What it is: The most respected academic framework for evaluating AI models holistically — accuracy, robustness, fairness, and more.
Why it matters for you: Vendor benchmarks flatter the vendor. HELM gives us an independent basis to judge which model actually fits your use case.
→ crfm.stanford.edu/helm

Epoch AI

What it is: Independent, data-led analysis of compute, cost, and capability trends in AI.
Why it matters for you: It grounds the cost and timing side of any AI business case in evidence rather than optimism.
→ epoch.ai

Semantic Scholar (Allen Institute for AI)

What it is: A free, machine-readable map of scholarly AI research and its citations.
Why it matters for you: It lets us verify claims to their primary source — the discipline behind every recommendation we make.
→ semanticscholar.org

Elements of AI (University of Helsinki)

What it is: A free, Nordic-built course that has introduced AI fundamentals to over a million people.
Why it matters for you: A trusted starting point we recommend for building shared AI literacy across a leadership team or workforce.
→ elementsofai.com

Open foundations — d2l.ai, Sutton & Barto, university courses

What it is: Canonical, open-access textbooks and curricula (Dive into Deep Learning; Reinforcement Learning by Sutton & Barto; Stanford and MIT open courses).
Why it matters for you: The basis of the capability-building work we deliver — and proof that strong AI foundations don't require licensing fees.
→ d2l.ai