XactAI reached approximately 7,000 licensees in Verisk's second quarter, an increase of nearly tenfold since March, alongside $806 million of revenue, up 4.3%, and free cash flow of $298 million, up 57.9% (Verisk earnings release, July 2026).
Three months earlier the same company's earnings call described governance friction stretching AI contracting timelines. One quarter later the friction is out of the script and the adoption curve has gone vertical.
Key Takeaways
- Approximately 7,000 XactAI licensees, nearly tenfold since March, with no matching revenue inflection: $806 million against $783 million in the first quarter, the same mid-single-digit trajectory.
- Adoption rides inside existing subscriptions. The two new Claude connectors are included in base subscriptions, with customers separately bearing the Claude token costs.
- A feature release inside a signed subscription triggers no procurement event, and the vendor-review provisions carriers wrote into their AI governance policies trigger on procurement events.
- Token consumption is a variable cost scaling with claim volume, which spikes in the catastrophe quarters when loss adjustment expense is already elevated.
- Free cash flow of $298 million, up 57.9%, funded a $200 million accelerated repurchase inside $1.9 billion of first-half buybacks, with full-year guidance reaffirmed at $3.19 to $3.24 billion.
Tenfold Licensees, Flat Revenue Line
The July 29 release describes a business performing to plan. Revenue of $806 million grew 4.3% as reported and 5.8% on an organic constant-currency basis, with adjusted EBITDA of $464 million, up 4.2%, holding a margin near 57.5%.
GAAP ran softer. Net income of $229 million fell 9.8% on a higher 24.6% effective tax rate, elevated interest expense and litigation-related legal fees, taking diluted GAAP EPS down 3.3% to $1.75 while adjusted EPS rose 5.3% to $1.98. Cash is where the strength concentrates: operating cash flow climbed 49.7% to $366 million and free cash flow reached $298 million, funding a $200 million accelerated repurchase that delivered an initial 949,190 shares, part of $1.9 billion of first-half buybacks with $800 million of authorization remaining (Yahoo Finance, July 2026).
Guidance was reaffirmed across the board: revenue of $3.19 to $3.24 billion, adjusted EBITDA of $1.79 to $1.83 billion, adjusted EPS of $7.45 to $7.75. For anyone reading vendor results as a proxy for carrier technology budgets, an unchanged outlook from the industry's central data utility says insurer spending on data and analytics is holding through the soft market.
The context that makes the licensee number notable is what Verisk said in April. The first-quarter call introduced governance-friction language absent from prior disclosures: intellectual property, privacy and compliance negotiations were extending contracting timelines for its most advanced analytics products. One quarter later the flagship claims AI grew from roughly one-tenth of its current base to about 7,000 licensees, described as a shift from experimentation to production scale (Investing.com transcript, July 2026).
Sequentially the financials barely moved. First-quarter revenue was $783 million with 7% subscription growth, so $806 million continues the same trajectory rather than reflecting an AI revenue surge. The tenfold expansion has not shown up as revenue because the adoption rides inside existing subscriptions, and Verisk's stated future options, subscription, transactional and agentic licensing, describe when the pricing conversation arrives instead.
Bundling, and the Two Costs It Moves
The Q2 disclosures show how the contracting friction cleared. The two new Claude connectors, which make Verisk underwriting and claims content queryable in natural language, are included in base subscriptions, with customers separately bearing the Claude token costs. XactAI's functions, claim summarization, photo labeling, document-data extraction and estimating recommendations, ride the Xactimate footprint adjusters already use.
Whatever the intent, the effect is that a new-product procurement, with its IP and privacy negotiation, becomes a feature release under an agreement already signed. That is a distribution result rather than a negotiation one, and every vendor watching this quarter's adoption number has now seen it work.
The first cost it moves is a governance one, running the opposite direction. Vendor-due-diligence expectations in the NAIC's AI model bulletin, and the vendor-review provisions carriers wrote into their own AI policies over the past two years, are triggered by procurement events: a new contract, a new tool, a new data flow. A capability arriving as a feature release inside an existing subscription crosses none of those tripwires. An adjuster whose estimating screen began offering AI-generated recommendations this spring did not wait on a model governance committee, because no event fired that would have convened one.
Verisk publishes documentation and has an established governance relationship with its regulator-facing content, which makes it the easy case rather than the hard one. The precedent is what carries, since the next embedded AI feature arrives from a vendor with thinner documentation and the same tripwires stay unarmed.
The second cost is metered. Connectors in the base subscription with token costs on the carrier is a consumption model: the marginal cost of every natural-language query, every summarized claim, every labeled photo lands on the carrier's expense line, metered by a frontier-model provider whose rates the carrier does not set. It is the same architecture visible in Palantir's second quarter, where usage-based platform economics grow with the customer's own adoption.
A claims organization rolling that tooling across thousands of adjusters has committed to a variable cost scaling with claim volume, the quantity that spikes in catastrophe quarters when loss adjustment expense is already elevated. Whether the combined ratio sees it, and whether rate indications capture it, depends on whether the consumption is allocated to LAE, to general technology expense, or left unallocated. A top-10 carrier is already live on the connectors.
What 7,000 AI-Assisted Estimators Do to Everyone's Benchmarks
The second-order effect lands in the data actuaries consume rather than in the expense line. Xactimate sits underneath a large share of US property claim estimates, which makes its outputs raw material for severity benchmarks, trend selections and reserving diagnostics across the industry.
When roughly 7,000 licensed users receive AI-generated estimating recommendations inside that workflow, the estimates stop being fully independent observations. Recommendation engines compress variance: adjusters accept suggested line items more often than they overrule them, which narrows the distribution of estimates for similar damage and can shift its center in whichever direction the training data leans.
None of that is visible in the aggregate severity statistics a pricing actuary downloads. The data simply becomes smoother and more internally correlated, which is the property that makes a benchmark look more reliable at exactly the moment it has become less so.
So property severity trend that moderates in 2026 carries at least three candidate explanations rather than one: genuine cost moderation, mix, and the arrival of recommendation-shaped estimates at scale. Separating them requires knowing what share of the underlying estimates ran through AI assistance, which is a disclosure that currently sits with vendors and third-party administrators rather than in the benchmark itself.
The profession spent a decade learning to adjust industry medical severity data for the effects of managed care. AI-assisted estimating is the property-lines version of the same instrumentation change, and it is arriving considerably faster than that one did.
Further Reading
- Verisk's Q1 2026 AI governance sales friction – the quarter this one reverses.
- Verisk's MCP connectors embed insurance analytics in Claude – the product launch behind the base-subscription bundling.
- ISO data in Claude and the actuary governance gap – what natural-language access to rating content means for filings.
- Palantir's Q2 and the carrier pricing-power question – the same consumption-cost architecture, one layer down the stack.
- Verisk's seven AI modules and the carrier pipeline – the product set now reaching production scale.