Akur8's Matrisk Acquisition Builds the First End-to-End Actuarial AI Platform

Akur8 has bought three actuarial software companies in fifteen months. The Matrisk acquisition in January 2026 sits between the purchase of Milliman's Arius reserving platform and the March 2026 acquisition of Slope Software.

The result is the first vendor to hold pricing, reserving, filings intelligence and life modelling in one platform. The Matrisk piece is the one that changes an actuarial workflow rather than extending a product line.

Key Takeaways

• Three acquisitions in fifteen months: Arius for P&C reserving, Matrisk for LLM search over SERFF rate filings, Slope Software for life and annuity cash-flow modelling.
• Over 50% ARR growth in 2025 on 50 new carrier clients, 25 or more of them in North America, against a base of more than 300 customers in 40-plus countries and over 3,000 actuaries using the platform daily.
• $180 million raised in total, including a $120 million Series C in September 2024, which makes Akur8 the best-capitalized pure-play actuarial AI vendor in the market.
• One to three weeks is what the manual competitive-intelligence phase typically adds to a filing preparation cycle. Discover is aimed squarely at that block of time.
• 150 insurance and consulting clients and over 1,500 U.S. users came with Arius, all of whom face a cloud migration to ReservingOne in the second half of 2026.

Three Acquisitions in Fifteen Months

Each deal fills a gap the original pricing engine could not close organically, and the original engine is worth stating first: Akur8 automates GLM and GAM construction with machine learning while keeping the output a standard filable model rather than a black box.

Arius brought P&C reserving, with 150 insurance and consulting clients and more than 1,500 U.S. users on triangle-based methods and stochastic modelling. Akur8 is rebuilding it as ReservingOne, cloud-native, for a second-half 2026 launch.

Matrisk brought a retrieval-augmented system that ingests U.S. P&C rate and rule filings from SERFF, normalizes them by carrier, state and line, and answers natural language queries against the corpus. It ships as Akur8 Discover, covering every P&C line and state with biweekly refresh and source-linked citations on every answer.

Slope Software, founded in Atlanta in 2015, brought hosted cash-flow modelling across pricing, valuation and forecasting for life carriers and reinsurers. It becomes Akur8 Life.

Chief Client Officer Brune de Linares framed the demand behind all three: "Insurers need to move faster as new pricing inputs, regulatory updates and market intelligence become available. At the same time, auditability, documentation and regulatory readiness are now baseline expectations."

Where Discover Lands in the Rate Review

The filings module attacks a specific, measurable block of the pricing cycle rather than the modelling itself.

Competitive intelligence currently means downloading PDFs from state department sites, parsing rate indication exhibits and building comparison spreadsheets by hand. That research phase typically adds one to three weeks to a filing preparation cycle, and it runs separately from model construction. Discover collapses it into a query: which carriers filed homeowners increases above 10% in Florida this quarter, what non-ISO variables competitors use for personal auto in Texas, which territory-factor justifications regulators have accepted in California.

Sitting behind it is the reason data preparation and documentation, not modelling, is where the actuarial hours go. Those steps run roughly 40% to 60% of the total time on a rate review, which is the block Akur8's agentic layer targets: ingest an extract, flag quality issues, suggest transformations, iterate specifications, document each decision, prepare the output for filing, with the actuary holding decision authority at each step.

Transparency is what makes the output filable. Because the engine emits standard GLM and GAM structures, validation runs on conventional methods, coefficient reasonableness, residual analysis, out-of-sample testing, rather than on an explanation exercise. That is the concrete separation from Earnix, whose optimization recommendations are harder to justify in states with strict rate justification requirements, and the reason it matters more in California, New York and New Jersey than elsewhere.

Factor	Akur8	Verisk	Guidewire	Earnix
Core value proposition	Transparent ML pricing with filings intelligence and reserving	Industry data and analytics foundation	Core-system-integrated pricing	Dynamic pricing optimization and rating
Core system dependency	Agnostic (any core system)	Data layer (supplements any platform)	Guidewire ecosystem	Agnostic (any core system)
Model transparency	Native (GLM/GAM output)	Varies by product	Moderate	Optimization can be opaque
Competitive intelligence	Built-in (Discover module)	Through data products	Not included	Not included
Reserving integration	ReservingOne (launching H2 2026)	Separate products	Not included	Not included
Life & annuity coverage	Akur8 Life (via Slope)	Limited	P&C only	Cross-line
Best fit	Mid-market carriers wanting a unified actuarial workflow	Carriers building custom stacks on Verisk data	Guidewire-native carriers	Carriers prioritizing real-time pricing optimization

Against Guidewire's PricingCenter the difference is core-system dependency: Akur8 runs on any core system, which is what makes it reachable for mid-market carriers not contemplating a core replacement. Against Verisk the relationship is closer to complementary, Verisk data feeding an Akur8 model, until Verisk's own analytics layer starts overlapping the pricing workflow.

The Filings Corpus Creates a Rate Development Question

Discover works by making systematic competitor analysis cheap. That is the product, and it is also where the second-order problem sits.

States with independent-filing requirements expect a carrier's indication to be developed from its own experience. A tool that surfaces what competitors filed, what justifications regulators accepted, and where the market is moving, refreshed biweekly across every state and line, invites the question of whether the resulting rate was independently developed or followed. The distinction that matters in a validation review is whether competitive data influenced model structure, which is defensible, or set the rate level, which is not, and Discover makes both equally easy.

That is the same governance surface the NAIC flagged when it named agentic AI insurance's next governance gap. Akur8's transparent output gives it a stronger starting position than an opaque optimizer, because the influence of any single input on the filed model is inspectable. It does not answer the question on its own.

Two execution risks sit alongside it. Three integrations are running at once: Arius to ReservingOne, Matrisk to Discover, Slope to Akur8 Life. Reserving actuaries carried over from Arius have a specific exposure there, since the cloud rebuild has to preserve the triangle methods and stochastic capabilities their opinions rely on.

And the agentic workflow is described, not deployed. The gap between an architecture that keeps governance front and center and an agent that reliably handles data quality checks without introducing errors an actuary must catch is the whole question, and it is not answerable from a roadmap. Carriers evaluating the platform, as we noted on carrier AI architecture decisions generally, should assess each module on what is in production rather than on the platform story.