Akur8

Akur8 is an AI-native actuarial platform headquartered in Paris, founded 2018 after four years of R&D. Offices in London, Madrid, Milan, New York, Atlanta, Montreal, and Tokyo reflect the company's bet that the rating-and-pricing layer is a global category rather than a regional one.

Funding. Approximately $180M raised cumulatively. The September 2024 Series C of $120M was led by One Peak with Partners Group, and — structurally interesting — participation from Guidewire Software (NYSE: GWRE). Guidewire invested from the incumbent modern rung into its AI-native successor; the two companies are now strategic partners at the boundary between the PAS and the rating engine. The round valued Akur8 at $400M.

Recent expansion. In 2024 Akur8 acquired the Arius reserving platform from Milliman, extending the product from pricing into reserving. Pricing and reserving are adjacent actuarial workflows that carriers historically bought from different vendors; the acquisition signals an ambition to own the full actuarial surface.

Where it sits in old → new → AI. The legacy rung is actuarial notebooks in SAS and R. Earnix defined the modern rung with a real-time pricing engine deployed alongside the PAS. Akur8 is the AI-native rung, automating the model-building loop itself. The Guidewire strategic investment is the clearest market signal that the three rungs are understood by the incumbent.

Common questions

What does Akur8 do? Akur8 automates the actuarial model-building process that insurers used to do by hand in SAS or R. Actuaries connect their claims data, let the platform train machine-learning pricing models, and get output they can send to regulators. The goal is to cut weeks of manual GLM work down to days while keeping the actuary in control of every decision.

How is Akur8 different from a traditional rating engine? A traditional rating engine executes a price formula that an actuary already built elsewhere. Akur8 builds the formula — it handles feature selection, model fitting, and validation — and then outputs a model the rating engine can consume. The two tools are complementary rather than competing; Akur8 sits upstream and hands off to platforms like Guidewire or Duck Creek.

Which insurers use Akur8? The customer list spans Europe, North America, and Asia. Named clients include Domestic & General and Co-operators in Canada, Link4 in Poland, and a long list of US carriers — Amica, Assurant, Arch Insurance Group, Heritage Insurance, NEXT Insurance, and Auto Club Group (AAA Michigan), among others. Sources: Akur8 customers page and 2024 annual report (akur8.com, accessed April 2026).

Why "transparent" ML — what does that mean here? Most ML models are black boxes: they produce a number but cannot explain which variables drove it. Akur8 uses generalized additive models (GAMs) and similar constrained architectures that produce smooth, monotonic curves regulators can inspect. An actuary can show a state insurance department exactly how age, vehicle type, or claim history shifts the predicted loss cost — which is a filing requirement in most markets.

Is Akur8 used outside Europe? Yes, and the US is now its largest growth market. Akur8 opened an Atlanta office in 2024 and its customer roster lists more than 20 US carriers, including regional mutuals (Western Reserve Group, CompSource Mutual), specialty insurers (James River, Canal Insurance), and larger groups (Assurant, Auto Club Group). The September 2024 Series C was explicitly aimed at accelerating US expansion (GlobeNewswire, September 2024).