Insurance · Deep dive

hyperexponential

London-founded pricing decision intelligence platform for specialty and commercial (re)insurers — hx Renew runs over $75B of annual commercial P&C premium at 40+ carriers including Beazley, Convex, Aviva, Allianz and Sompo, on $91M raised through a $73M Series B led by Battery Ventures (January 2024), and is now pivoting from a pricing tool into an 'agentic underwriting workbench' with the July 2026 launch of hyperoperator.

emerging

The question that decides it: Can hyperexponential unseat WTW Radar's decades of incumbency across Lloyd's and specialty pricing before Akur8 verticalizes into specialty from its personal- and commercial-lines GLM base, or before Guidewire and Duck Creek make pricing a native, bundled module of the policy-admin systems carriers already pay for? Falsifiable test for winning: 5+ top-20 Lloyd's syndicates provably displace Radar as primary pricing engine, disclosed ARR crosses $100M, and the US commercial P&C customer count clears 20+. Falsifiable test for losing: Akur8 lands 3+ Lloyd's syndicates first, Guidewire or Duck Creek ships a genuinely competitive native pricing/agentic-underwriting engine, or hx's YoY growth decelerates below 40%.

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

hyperexponential (hx) is a London-founded software company selling pricing decision intelligence to specialty and commercial (re)insurers — the actuaries and underwriters who price marine, cyber, professional indemnity, D&O, energy and aviation risk at Lloyd’s syndicates and specialty carriers worldwide. Its flagship product, hx Renew, is a browser-based, Python-native platform that lets actuaries build, test, deploy and monitor pricing models at something close to the speed of Excel, while giving the carrier’s IT and compliance functions the governance, version control and auditability Excel never had. As of 2026 the company says hx Renew underpins pricing decisions across more than $75B of annual commercial P&C premium at 40-plus carriers, including Aviva, Allianz, Sompo, Beazley, Markel, HDI, Cincinnati, Convex and Bowhead (hyperexponential.com, 2026). Founded in 2017, hx has raised roughly $91M in disclosed venture funding — an $18M Series A led by Highland Europe (June 2021) and a $73M Series B led by Battery Ventures with Andreessen Horowitz (January 2024) — and employs roughly 204 people (LeadIQ, 2026). The company matters now because it just repositioned itself: in July 2026 it launched hyperoperator, an AI agent that claims to take a broker submission from raw email to a fully priced, quoted risk in under three minutes, pushing hx from a pricing tool into what it now calls an “agentic underwriting workbench.”

Founding story

Amrit Santhirasenan started his career in 2005 as an actuary at Catlin, the Lloyd’s (re)insurer, where at 25 he was seconded to Toronto to build Catlin’s Canadian actuarial function from nothing. He later moved to Tokio Marine Kiln as Head of Pricing and Analytics, where he built the firm’s first-ever technical pricing team — and ran directly into the problem hx was built to solve. Actuarial pricing models for specialty lines are genuinely hard: heavy-tailed loss distributions, thin data, bespoke risk factors per line of business, and constant iteration as underwriters push back on rates. The tooling available to do that work in 2015-2017 was Excel, VBA macros, and — at the high end — WTW’s decades-old Radar/Emblem stack, none of which gave actuaries fast iteration, version control, or a clean handoff to underwriters and IT. Santhirasenan has described founding hx in 2017 out of frustration that no insurance-specific platform existed to bring actuaries, underwriters and the industry’s exponentially growing datasets together in one place (Insurance Times, 2023; InsTech, 2023) — a frustration Michael Johnson shared directly: Johnson had followed a nearly identical path, from actuarial roles at Catlin (2011-2015) into the same Pricing & Analytics function at Tokio Marine Kiln (2015-2017) that Santhirasenan led. The two co-founded hyperexponential in London in November 2017. hx’s first real market validation came in September 2019, when it joined the third cohort of the Lloyd’s Lab accelerator — a ten-week programme through which it reportedly built a working actuarial pricing model in 32 hours, roughly ten times faster than the industry’s typical build cycle, and got its first direct line into Lloyd’s syndicate buyers.

How it works

The mechanical case against Excel-based pricing is specific: a pricing actuary at a Lloyd’s syndicate typically maintains dozens of line-of-business rating models as linked spreadsheets with embedded VBA, each one a single point of failure, with no real version control, no automated regression testing when a rate change is pushed, and no clean audit trail for the syndicate’s own risk and compliance functions or for Lloyd’s oversight. hx Renew replaces that stack with a browser-based environment built around a Python runtime, so actuaries write and iterate models in a language built for statistical work rather than spreadsheet formulas, while the platform layers in Git-style version control, automated batch testing across historical bordereaux, “what-if” scenario analysis, and automated regression testing that flags calculation errors before a rate change reaches production. The workflow hx sells runs: data ingestion from bordereaux, exposure and claims systems → model build and refit (GLM/GAM and increasingly ML-based approaches) inside the Python environment → underwriter overlay, where the underwriter’s own judgment adjusts the actuarial indication before a quote is issued → real-time quote and bind, pushed into the carrier’s policy-admin system via API → ongoing model-performance monitoring, tracking realized loss ratios against what the model predicted. hx’s newest layer, hyperoperator (launched July 2026), sits upstream of all of that: it is an agent that reads an incoming broker submission email, extracts and structures the risk data, triages it against the carrier’s stated appetite, and pushes a decision-ready, priced file back to the underwriter — with four operating modes (straight-through execution on routine business, human-in-the-loop review, ad hoc underwriting analysis, and always-on portfolio monitoring) so carriers can choose how much of the loop to automate.

Product and business overview

hx Renew is the core pricing platform: model build, deploy, monitor, and the API layer connecting into policy administration systems including Guidewire, Duck Creek and Sapiens. hyperoperator, launched July 2026, is the newer agentic layer sitting on top of Renew’s pricing models, automating submission intake and triage through to a bindable, priced quote inside the carrier’s own authority and appetite controls; it entered controlled customer rollout in the second half of 2026. hx does not appear to sell separately branded “hx Insight” or “hx Studio” products under those names — the public product architecture as of 2026 is Renew (pricing) plus hyperoperator (agentic underwriting), both marketed together as a single “agentic underwriting workbench.” The company also runs a formal partner ecosystem: a 2023 partnership with Akur8 lets carriers pair hx’s specialty/commercial pricing workflow with Akur8’s automated GLM/GAM model generation, and productized REST APIs connect hx into Guidewire, Duck Creek, Sapiens and Milliman systems so pricing outputs flow directly into a carrier’s existing policy-admin stack rather than requiring a rip-and-replace.

Business model and pricing

hx does not publish a public price list; deals are enterprise annual SaaS contracts negotiated per carrier, typically scoped by number of lines of business, number of actuarial/underwriting users, and which modules (Renew alone, versus Renew plus hyperoperator) are included. Trade coverage of Lloyd’s-market pricing-technology deals generally puts syndicate-level deployments in the low-to-high six figures of annual contract value, consistent with the scale of the customer base hx discloses (40+ carriers generating a combined $75B+ of underlying premium, implying an average customer size well above a small MGA). Revenue is not broken out publicly. Latka’s independently compiled estimate puts hx at roughly $22.4M ARR in 2025 — modest against $91M of disclosed venture capital, though the number should be read as a third-party estimate rather than a company disclosure, and it likely understates 2026 revenue given the hyperoperator launch and the reported jump in premium running through the platform from $45B (January 2024) to $75B+ (2026).

Traction over time

The trajectory is a genuine but unspectacular climb in disclosed metrics (headcount roughly 50 → 160 → 204 across 2021-2026; premium under management $45B → $75B+ in about two years) punctuated by one clear strategic pivot — the move from “pricing platform” to “agentic underwriting workbench” in mid-2026, which is as much a repositioning against AI-underwriting competitors like Cytora and Sixfold as it is organic product evolution.

Market analysis

Lloyd’s of London alone wrote £57.9B (roughly $73B) of gross written premium in 2025, up 4.2% year-on-year, and Lloyd’s estimates it represents about 10% of global insurance and reinsurance business (Lloyd’s full-year results, 2026) — giving a rough sense of scale for the specialty/commercial segment hx targets globally, which runs well into the hundreds of billions of dollars in GWP once US excess-and-surplus lines, European specialty and global reinsurance are included. The narrower software market hx actually sells into is far smaller: independent market-research estimates put the global insurance pricing software market at roughly $2.34B in 2024, growing at a 10.2% CAGR to about $5.54B by 2033, while a narrower “actuarial software for insurance pricing” category is sized at $587M-$891M in 2024-2025 growing to $860M-$1.4B by 2030-2032 at roughly 6.5-6.8% CAGR (QY Research, datainsightsmarket, 2025-2026). hx’s own addressable slice sits between those figures — the specialty/commercial actuarial-pricing segment specifically, which is smaller still but where hx claims category leadership. Two structural tailwinds matter: continued Lloyd’s/London-market pressure to modernize away from Excel-based pricing (a theme Lloyd’s itself pushes through its Pricing Maturity Matrix guidance, which hx has co-branded content around), and the broader 2025-2026 push toward agentic AI in underwriting, which is pulling hx, Cytora, Sixfold and Concirrus all toward overlapping “submission-to-bind” product territory rather than staying in separate lanes.

Competitive intel

WTW Radar/Radar Live is the incumbent hx was built to displace — three decades of development, dominant in personal lines and deeply embedded across Lloyd’s syndicates and specialty carriers, now shipping Radar 5 with generative-AI features that close part of the modernization gap hx’s whole pitch depends on. Akur8 (~$180M raised, 330+ carriers) is simultaneously hx’s partner and its most credible future competitor: the 2023 partnership divides labor — Akur8 automates GLM/GAM model generation, hx owns the specialty/commercial underwriting workflow — but Akur8 has explicitly signaled ambitions to move upmarket into specialty and commercial lines, which would put the two head-to-head. Earnix ($150M+ raised, Insight Partners-backed) dominates high-volume personal-lines real-time rating at carriers like Allstate and Munich Re; hx’s own marketing argues Earnix lacks the underwriter-facing tooling complex specialty risk requires. Verisk/ISO sells pre-built rating plans that let smaller US carriers skip hiring pricing actuaries entirely — a cheap substitute that caps hx’s small-carrier TAM. Milliman competes for the same chief-actuary budget through outsourced actuarial consulting and its (now Akur8-owned) Arius reserving software. Concirrus (Inspire) and Cytora (Autopilot) are London-market AI-underwriting platforms whose product scope — marine/specialty telemetry for Concirrus, commercial submission triage and straight-through processing for Cytora — increasingly overlaps hx’s new hyperoperator ambitions rather than staying purely adjacent. The most structural threat is Guidewire and Duck Creek shipping native rating engines and AI-assisted underwriting features directly into the policy-admin suites carriers already pay for and already run hx’s own APIs against — the same bundling risk Akur8 lives with as a Guidewire portfolio company.

History and evolution

• Nov 2017 — Founded in London by Amrit Santhirasenan and Michael Johnson.
• Sep 2019 — Joins Lloyd’s Lab accelerator cohort 3; builds a demo pricing model in 32 hours.
• Jun 2021 — $18M Series A led by Highland Europe.
• End 2023 — ~160 employees across three offices (InsTech/Insurance Times).
• 2023 — Partnership with Akur8 announced, pairing GLM/GAM automation with hx’s specialty/commercial workflow; productized API integrations with Guidewire, Duck Creek, Sapiens and Milliman.
• 11 Jan 2024 — $73M Series B led by Battery Ventures with a16z and Highland Europe; Marcus Ryu (Battery Ventures partner, Guidewire co-founder/former CEO) joins the board; company announces New York office and a plan to more than double headcount past 200; discloses $45B+ of annual premium running through hx Renew.
• 2025 — Latka estimates ~$22.4M ARR.
• Mar 2026 — Partnership with Banyan Risk marks first deployment of hx’s full agentic underwriting suite across US, UK, Canada and Bermuda.
• 8 Jul 2026 — Launches hyperoperator, an agentic underwriting AI taking broker submissions to priced, decision-ready files; repositions the company as an “agentic underwriting workbench”; discloses 40+ carriers and $75B+ of annual premium under management.
• Sep 2026 — Publishes industry survey finding underwriters cite losing senior judgment as their biggest AI-era fear, even as firms underinvest in preserving it — messaging that doubles as hyperoperator go-to-market.

What people say

The case for. Customer case studies read consistently on one theme: speed of model iteration. AEGIS London — a top-quartile Lloyd’s syndicate writing over $1B GWP in 2024 that had been running 58 individual pricing models on a bespoke internal web platform — is cited by hx as a reference customer for consolidating that sprawl onto Renew. Convex, hx’s longest-tenured customer, is cited for using hx to build its entire pricing-tool suite while hitting faster speed-to-market. On Glassdoor, hx scores 3.9 out of 5 across 69 reviews with 71% of employees saying they’d recommend the company to a friend; reviewers praise a genuinely collaborative, knowledge-sharing culture where senior people are accessible and the mission feels shared (Glassdoor, 2026).

The complaints. The same Glassdoor reviews that praise the culture also describe it as intense and demanding: startup-typical long hours, a fast-moving and somewhat unstructured environment that “works best for self-driven people comfortable with ambiguity” but can feel disorienting for others, and at least one review describing the culture as “more on paper” than in daily practice, with employee events landing awkwardly and in-office collaboration falling short of the pitch. Independently, hx has essentially no visible footprint on G2, Capterra or TrustRadius — the standard enterprise-software review venues — which is itself a signal: buyers doing procurement diligence on hx get customer references hx selects, not an independent peer-review layer, the same structural gap that shows up at close competitor Akur8. On product substance, the clearest external critique comes from WTW’s own competitive position rather than any single review: Radar’s three decades of Lloyd’s/London-market entrenchment means most RFPs still start from a Radar baseline, and hx’s own comparison content implicitly concedes this by spending as much time arguing why switching is worth the disruption as it does describing hx’s own features.

Outlook: the open question

Can hyperexponential unseat WTW Radar’s Lloyd’s/specialty incumbency before Akur8 verticalizes into specialty, or before Guidewire/Duck Creek bundle competitive pricing natively? The bull case rests on three real facts: hx has grown premium-under-management from $45B to $75B+ in about two years, it has a governance-grade board addition in Marcus Ryu (who built exactly this kind of platform-versus-incumbent fight once already at Guidewire against legacy mainframe policy administration), and it moved first into agentic underwriting with hyperoperator ahead of Akur8, Earnix and most of the Lloyd’s-market AI-underwriting cohort. Winning looks like: 5+ top-20 Lloyd’s syndicates provably displacing Radar as their primary pricing engine rather than running hx alongside it, disclosed ARR clearing $100M, and a US commercial P&C customer count above 20 that shows the New York office converting into real revenue rather than a headline. Losing looks like: Akur8 announcing 3+ Lloyd’s syndicate wins before hx locks down its next tranche of specialty accounts, Guidewire or Duck Creek shipping an in-suite pricing/agentic-underwriting module good enough that carriers stop paying for a standalone layer, or hx’s own growth decelerating below 40% YoY — which an estimated $22M ARR on $91M raised suggests is not yet guaranteed. The most honest read of hyperoperator’s July 2026 launch is that it is a hedge as much as an advance: hx is racing to own the agentic-underwriting layer before Cytora, Sixfold or a Guidewire-native feature owns it instead, which means the open question is really two open questions stacked on top of each other.

How to attack it

Do not attack hx on pure pricing-model tooling for Lloyd’s specialty lines — the workflow trust required (actuaries, underwriters, compliance and IT all signing off on a rating-model change-management process) took hx nine years and $91M to build, and every incumbent RFP still starts from a Radar or hx baseline rather than a blank page. Attack the seams hx’s own July 2026 pivot exposes.

Wedge 1: An open-source, Pydantic-native pricing runtime. hx’s core defensibility is a proprietary Python execution environment with built-in governance. A well-funded team could ship an open-source, schema-validated (Pydantic-style) pricing runtime that any carrier’s own actuaries can self-host and audit line-by-line — turning hx’s “enterprise-grade governance” pitch into a commodity and competing purely on deployment speed and support, the same way open-source infrastructure has hollowed out proprietary middleware categories before.

Wedge 2: Agentic underwriting that starts from the broker submission, not the carrier’s pricing model. hyperoperator still assumes a carrier has already built its Renew pricing models; a leaner attacker could build an AI underwriting agent that works against any carrier’s existing rating engine (Radar, hx, Akur8, or a spreadsheet) via API, monetizing the submission-to-bind layer without needing to first win the pricing-model deal. This directly targets carriers unwilling to rip out Radar but willing to add an AI layer on top.

Wedge 3: Guidewire/Duck Creek bundling. Both platforms already have hx’s own API integrations built for them; either vendor shipping a genuinely competitive native pricing and agentic-underwriting module inside the suite carriers already pay license fees for removes the economic case for a third-party layer entirely. This is the single biggest structural risk to hx and the easiest wedge for a well-capitalized platform play to execute, because the integration and distribution work is already done.

Enumerated weaknesses. (i) Heavy deployment and integration overhead — every hx implementation touches bordereaux ingestion, model migration and policy-admin API work that keeps sales cycles long and services-heavy. (ii) Actuary hiring is a bottleneck on both sides of the table — hx needs domain-expert implementation staff to sell into an actuary-run buying process, capping how fast it can scale logos. (iii) WTW Radar retains an estimated multi-decade mindshare advantage in Lloyd’s RFPs that no product comparison page fully offsets. (iv) No visible G2/Capterra review layer leaves enterprise buyers without independent validation. (v) hyperoperator enters “controlled rollout” in 2026 against Cytora and Sixfold, both of which have been shipping submission-triage AI longer.

Adjacent-segment play

The most direct adjacency is personal auto and home pricing, competing head-on with Akur8 and Earnix on their home turf. hx’s Python-native, version-controlled model architecture generalizes technically to any GLM/GAM pricing workflow — the reason the Akur8 partnership exists in the first place — but the personal-lines buying process (procurement-led, price-sensitive, dominated by Radar and Earnix incumbency) is a worse fit for hx’s high-touch, actuary-relationship-led go-to-market than specialty commercial ever was, so this is a real option but not an easy one.

A cleaner adjacency is reinsurance retrocession pricing — treaty-layer, cession and cat-bond pricing uses the same severity/frequency modeling core as primary specialty pricing but with contract structures (excess-of-loss layers, reinstatement provisions) neither Akur8 nor most primary-pricing platforms natively model; hx already touches this market through Lloyd’s reinsurance syndicates and could extend Renew’s model library rather than build new infrastructure.

A third is MGA-in-a-box pricing infrastructure for the wave of digitally native MGAs (Vouch, Coalition, Cover Genius) that need actuarial-grade pricing without hiring a full actuarial function — a smaller-ACV, higher-volume version of the same core product, sold on API access rather than a full enterprise deployment.

The adjacency that does not generalize is standalone cyber pricing as an isolated product: cyber pricing already has dedicated, better-capitalized specialists (Coalition, At-Bay) building proprietary telemetry-driven models that a general-purpose pricing platform cannot match without the same security-telemetry data asset those companies have spent years accumulating.

Sources and further reading

• hyperexponential Raises $73M Series B Funding — hyperexponential, 11 January 2024
• UK insurtech Hyperexponential raises $73M and eyes US expansion — TechCrunch, 11 January 2024
• Backing the Next Generation of Insurtech: Battery leads hyperexponential’s Series B — Battery Ventures, January 2024
• London-based insurtech hyperexponential closes $18M round led by Highland Europe — TechCrunch, 30 June 2021
• Amrit Santhirasenan: Hyperexponential growing organically amid AI revolution — Insurance Times, 2023
• Modelling Reimagined – Amrit Santhirasenan, hyperexponential — InsTech, 2023
• hyperexponential Introduces hyperoperator, the Unified Agent for Underwriting Work — GlobeNewswire, 8 July 2026
• WTW Radar vs hx platform: actuarial pricing platform comparison — hyperexponential blog, 2026
• hx vs Earnix vs Radar: Choosing the right underwriting platform — hyperexponential blog, 2026
• hyperexponential and Akur8 partner to power data-driven specialty and commercial pricing — hyperexponential, 2023
• Convex streamlined workflows and pricing decisions with hx — hyperexponential customer stories, 2026
• Lloyd’s market delivers strong full year performance; very strong balance sheet; increased capital — Lloyd’s, 2026
• hyperexponential Reviews (69) — Glassdoor — Glassdoor, 2026
• hyperexponential Revenue 2025: $22.4M ARR — Latka, 2025