CLARA puts an AI workforce on every casualty and workers’ comp claim

CLARA Analytics launched Agentic Intelligence, an end-to-end AI workforce built into its CLARAty.ai platform for complex casualty and workers’ compensation claims, moving from insight dashboards to agents that recommend and help carry out next steps while adjusters stay in control.

FINANCE desk — claims is where insurance P&L lives; putting agents inside the claim file (not another sidecar chatbot) is how carriers try to cut cycle time without giving up adjuster control.

What the page says the product is. Agentic Intelligence is an end-to-end AI workforce for complex casualty and workers’ compensation claims, built into the CLARAty.ai platform. A casualty claim, here, is a claim that someone else was hurt, or that their property was damaged. Workers’ compensation, which the page shortens to workers’ comp, pays when an employee is hurt at work. First notice of loss is the moment the insurer first hears a claim exists. An adjuster is the person at the insurer who works that claim and keeps the decision. An agent, here, is software that recommends a next step and helps carry it out, not only a number on a screen. The page says the product puts a team of those agents on every claim, from first notice of loss through resolution, while adjusters stay in control of every decision. Every claim, and around the clock, are the page’s words. This desk did not open a claim file.

The shift the page says it is making. The launch moves CLARA from a system that surfaces insights to one that recommends next steps and helps carry them out. The agents sit inside the claim file. The page calls that file the place where workers’ comp and casualty losses are decided. The same paragraph says the agents use a 10-year proprietary data moat of historical outcomes that competitors cannot replicate or license. A data moat, in this sentence, is the company’s name for a pile of past claim results it says rivals cannot copy or buy a right to use. Proprietary means the company says it owns that record. Ten years, and cannot replicate or license, are the company’s. This desk did not see the archive.

Two sentences from the chief executive, as the company’s, not as a result. Heather H. Wilson said the agents continuously evaluate how a claim is developing and catch exposure issues before they become a problem. Exposure, here, means how large the loss could get. She said Agentic Intelligence is a partner that reviews claims data and documents and surfaces real-time changes in severity, litigation potential, fraud risk, and closure opportunities. Severity is how serious or expensive the claim looks. Litigation potential is the chance it becomes a lawsuit. Fraud risk is the chance the claim is not honest. A closure opportunity is a chance to finish the claim. She added that CLARA has gone from a system of intelligence to a system of action by becoming a full claims agentic platform. Working together, she said, the agents pull the relevant information together, give transparent reasoning and adaptive intelligence, surface opportunities, and build the bridge from intelligence to action. The page prints her title as CEO. File the name, the title, and those sentences as the release’s. A quote is not a measured drop in what a claim costs.

Where a score comes from. The section the page titles “Inside Agentic Intelligence” says adjusters get instant access to where a score came from and how current the data under it is, so they are not chasing source documents to trust a number. A score, here, is a number the platform puts on a claim. The page says the agents add context and reasoning for claims professionals and for actuarial professionals. Actuarial work is the math side of insurance: what a set of claims is likely to cost. File that access as the company’s. This desk did not trace a score back to a document.

Five features, in the page’s order. The heading above them is “Transforming Claims Intelligence Through Agentic Reasoning.” Complete Lineage and Outcome “Ground Truth” is instant access to glossary definitions, who owns a metric, and which version is in use. Lineage is the path back to where a number came from. Ground truth, in the company’s quotation marks, is the outcome it treats as the real result. Data Freshness and Quality is a live look at whether the data is healthy and whether the source checks out. Cohort Contextualization is a drill-down into coverage groups, exclusions, and the groups of claims behind a number. A cohort is a group of similar claims. An exclusion is something the policy does not cover. Transparent Rationale is the scoring logic explained and tied back to the claim documents and the fields in them. Interactive Deep Dives is a conversational agent, a chat a person can question, for more context on the risk and on the recommended action, across structured and unstructured data. Structured data is the fields in a table. Unstructured data is the notes, letters, and other documents. File the five names as the release’s. This desk did not click through a claim.

The count, and whose count it is. The page says the product is powered by self-learning data pipelines and a 10-year data moat of over 7 million claims. A pipeline, here, is the path data takes into the platform. Self-learning is the company’s word for pipes that fill gaps as they run. Over 7 million is the company’s count of historical claims. It is the same 10-year moat as the earlier sentence, now with a number on it. The page says CLARA resolves missing data in real time and replaces rigid rules with dynamic querying. Dynamic querying means asking the data a question instead of running a fixed rule. It says the platform eliminates manual coding, which here means writing those rules by hand, so a team can slice the data and stress-test it sooner. It says that removes friction, builds the adjuster’s skill, and leaves people on the complex decisions. The section ends on one line: adjusters keep the decision, and CLARA owns the evidence trail behind it. Eliminates, and highly accurate, are the page’s words. Over 7 million is not a count this desk audited.

Where the company says it will show the product. CLARA says it will demonstrate the platform at ITC Vegas from Sept. 29 to Oct. 1, in Booth 2825, Bayside Halls D-F, and that it is a Gold Sponsor. On Sept. 30, from 2:05 to 2:35 p.m., CLARA and Nationwide take the stage for a session titled “Stop Leaving Claims Risk on the Table.” The page does not print a year or a time zone on those lines. The release date is 22 Sep 2026. Nationwide, on this page, is the other name on that session. The about box does not call Nationwide a customer. Do not turn a conference slot into a contract. Gold Sponsor is the company’s word. This desk is not attending.

The product chief, as color, not as a dollar figure. Mubbin Rabbani, chief product officer, said the company is not only delivering faster technology. He said it is delivering measurable, quantitative financial and business outcomes. He said insurers get turn-key access to an advanced AI workforce backed by robust guardrails and human validation. Turn-key is the page’s hyphenation. It means ready to switch on. A guardrail, here, is a limit meant to keep the software inside a rule. Human validation means a person still checks the work. The page does not print the dollar outcome, the list of guardrails, or the validation step. Do not invent them. File the name, the title, and those sentences as the release’s.

The about box, and where the boast stops. CLARA calls itself the leading provider of AI as a service, which the page shortens to AIaaS, for casualty claims. AI as a service means the customer uses the software. It does not mean the customer trains the model. The box says the product is for insurance carriers, managing general agents, reinsurers, and self-insured organizations. A managing general agent, shortened to MGA, is a firm allowed to write policies for a carrier. A reinsurer insures the insurer. A self-insured organization pays its own claims. The page says the platform uses image recognition, natural language processing, and other AI methods on medical notes, legal demand packages, bills, and other documents around a claim. Natural language processing is software that reads writing. Image recognition is software that reads pictures. The page says those insights are augmented intelligence that helps a claim professional cut claim cost. Augmented intelligence is the company’s phrase for help a person still directs. Customers, the box says, include companies from the top 25 global insurance carriers to large third-party administrators and self-insured organizations. A third-party administrator, often shortened to TPA, handles claims for someone else. From the top 25 is a range of customer types. It is not a named list, and it is not a claim that every one of those 25 is a customer. The company was founded in 2017. Headquarters are in California’s Silicon Valley. The page does not name a city. The site it prints is www.claraanalytics.com. It also says to follow the company on LinkedIn and @CLARAAnalytics. Leading, and the top-25 sentence, are the company’s. This desk did not call a carrier.

Plain English for the rest of the card: casualty claim = someone else was hurt, or their property was damaged. workers’ comp = pay when an employee is hurt at work. first notice of loss = the insurer first hears the claim exists. adjuster = the person who keeps the decision. agent = software that recommends a next step and helps carry it out. claim file = the record of that claim. data moat = the company’s name for past results it says rivals cannot copy or license. 10 years and over 7 million claims are the company’s figures for that moat. exposure = how large the loss could get. severity = how serious or expensive the claim looks. litigation potential = the chance of a lawsuit. fraud risk = the chance the claim is not honest. closure = a chance to finish the claim. score = a number the platform puts on a claim. actuarial = the math of what claims are likely to cost. lineage = where a number came from. ground truth = the outcome the company treats as the real result. cohort = a group of similar claims. exclusion = something the policy does not cover. structured data = fields in a table. unstructured data = notes and documents. pipeline = the path data takes in. dynamic querying = asking the data a question instead of a fixed rule. manual coding = writing those rules by hand. AIaaS = AI as a service. MGA = a firm that can write policies for a carrier. reinsurer = an insurer for insurers. self-insured = an organization that pays its own claims. TPA = a firm that handles claims for someone else. turn-key = ready to switch on. guardrail = a limit on the software. human validation = a person still checks the work. Booth 2825 = the ITC Vegas booth the page names. This filing is the 22 Sep launch. It is not a price, and it is not a customer list.

PRIMARY here: CLARA Analytics’ 22 Sep 2026 Business Wire release, “CLARA Analytics Unveils Industry’s First End-to-End Agentic Claims Intelligence Platform,” carried on FinancialContent at 10:12 AM EDT — Tier A PRIMARY, the company’s own record. The page this desk read is that syndication. It does not print a city, and it does not print a businesswire.com address. The launch inside CLARAty.ai, the every-claim and around-the-clock lines, the move from insights to recommended next steps, the claim-file placement, the 10-year moat, the Wilson title and quotes, the five feature names, the over-7-million-claims line, the self-learning pipelines, the adjusters-keep-the-decision line, the ITC Vegas dates, Booth 2825, the Nationwide session title, the Rabbani title and quotes, the 2017 founding, the Silicon Valley headquarters with no city, and the top-25 customer range are the release’s. NOT claimed: that this desk ranked a “first” or a “leading,” opened a claim file, counted 7 million claims, saw a license a rival could not buy, measured a cost cut, named the 25 carriers, treated Nationwide as a customer, found a price, a city, a stock tip, or investment advice. Distinct from the already-filed sela-21m, creatio-bank-ai-300m, and heidi-340m.