Swallow

Swallow sold its pricing platform for roughly two years before the AI line existed. Remove the models and a deterministic pricing engine, visual builder, test harness, approval workflow and rating interface all remain. AI now carries the throughput on two shipped lines. Herbie, described as the AI actuary inside the platform, builds a tested and deployable pricing model from a prompt and generates up to a hundred test cases from an instruction. The filings line makes twenty five years of United States rate filings executable, extracting rating logic from documents rather than summarizing them.

Oversight in this product depends on the path. On the filing path, a credentialed actuary, the customer's own or a partner, reviews, certifies and manages regulator objection responses. The split is quantified: eighty to ninety percent of artifacts are generated automatically and professional judgment is retained for the rest. That names and staffs an adjudication layer, places it in the path, and ties the reviewer to an external professional credential. The distribution path is different.

A voice agent runs renewal campaigns at scale, generates updated quotes in real time and is described as negotiating live, with no threshold, review structure or escalation route published. Deployment gates are real: test suites run automatically on any change, changes go through approvals, and one click rollback keeps an audit record of who changed what and why.

Pricing model governance is strong and the AI layer is silent. Every publish creates a versioned interface, and previous versions stay live and callable. One click rollback keeps an audit record of who changed what and why. Testing covers scenario and pivot testing, impact testing before production, what if analysis and back testing on historical data. Controlled experiments route a share of quotes between model versions.

Extraction accuracy is the open question, and it is central rather than incidental. The filings product reads rating logic out of documents, and a wrongly extracted factor table propagates silently into every quote. Reconciled example calculations are offered against a filing's own worked examples, which is the closest thing to an accuracy control. No tolerance, pass rate, failure handling or human review rate is published, and nothing describes the models behind Herbie.

Two of five named customers have full case studies with outcomes. Rivr launched complex life products in weeks rather than months, quoted by founder and chief executive Will Clarkson Webb. Open brought a premium United Kingdom home insurance product to market in under a week. The other named customers are MGT Insurance, Ouch and ISC. Named customers, quantified outcomes and a named executive speaking publicly make this solid evidence.

The scale claims are bigger but sit apart from the attributed ones. The customer serving three million quotes a month and the client above $1 billion of revenue are not identified, which limits how much the scale claims can carry on their own.

Nothing is published on whether customer quote data, pricing models or uploaded workbooks train anything. The vendor names the hallucination risk, then exempts itself with an unfalsifiable claim on the same page. One answer states that AI can hallucinate quote questions, quote prices and the rules around quotes. Another states that the vendor's own channels handle new quotes, amendments and recall via AI without hallucination.

On the positive side, the market analytics product builds peer comparison from filings that are public by law. It delivers what pooled benchmarking delivers elsewhere without holding any competitor's confidential data, because a public transparency regime makes confidential pooling unnecessary.

There is no position at all on GLBA, despite a product line built specifically for United States carriers. The company states the platform is built with regimes such as GDPR in mind and that data can be exported, deleted or audited at any time. It is registered with the United Kingdom data protection regulator and publishes a privacy policy. Built with a regime in mind is a gesture, not a commitment.

The export and delete capability is a product feature, not a promise about the vendor's own processing. No data processing agreement, subprocessor list or statement about what the vendor does with quote data is published.

ISO 27001 certification is stated plainly, but only in a buried support answer. Encryption in transit and at rest, role based permissions, audit logs and complete version history are also described. The company is registered with the United Kingdom data protection regulator. The footer badge reads only certified, with no numbered standard, so a real certification is missing from the badge row that exists to advertise it. There is no trust portal or certificate link, and no scope or audit period is stated. No SOC report of any kind is published for a vendor whose output feeds premium calculations.

Unlicensed and unsupervised, Swallow works close to the regulatory machinery without being inside it. The product indexes the filing portal every carrier uses and generates the actuarial memorandum, rate manual, exhibits and rule pages. It also answers regulator objections and encodes state specific rating restrictions. The party answerable to regulators remains the carrier and the certifying actuary. Compliance guidance across three regions is a product feature, not standing. Swallow writes filings for a living, but no state supervisor has examined or approved its own models.

No fairness testing, protected class position or governance disclosure of any kind sits beside a product that sells price optimization. The mechanism is described in detail. The platform routes live customers into controlled pricing experiments and scores the versions on sales, and the voice agent negotiates renewal prices in real time with individual consumers. Optimizing against propensity to accept rather than against risk is exactly the practice consumer fairness rules address. In this product it is not an inference from a product name but a described workflow with the success criterion stated.

A pricing error introduced between the filed document and the live quote has no named owner. No recourse route or allocation of responsibility is published beyond standard terms. The chain is long. An approved filing is machine extracted into an executable rater, and the rater is published as a live interface. A voice agent quotes and negotiates against it, and a consumer is priced. The certifying actuary signs the filing, not the extraction, and the carrier owns the customer relationship.

The only supplier named is peripheral: a direct integration with a speech synthesis provider for the voice channel. Nothing names the model doing actuarial work, reading rate filings or generating pricing models, which is where the buyer's exposure sits. Knowing the voice vendor tells a buyer nothing about the system reading its rate tables. No model inventory, default or version appears anywhere.

Swallow is built as a rating layer that runs independently of policy administration, sold as a way to move pricing out of a legacy core while leaving the core stable. Secure REST interfaces carry a versioned endpoint per publish, alongside webhooks, batch endpoints and software development kits. Embeddable forms, chat and voice channels sit on top, with exports to Metabase, Redshift, Athena, Snowflake, Google Sheets and Excel. It is listed on the Amazon Web Services marketplace.

The performance envelope is published: one customer serves more than three million quotes a month, and the platform handles price comparison website volumes at low latency and high concurrency. No named core system connector exists. Established rating vendors, by contrast, integrate with named policy suites.

Single tenant software as a service on Amazon Web Services with a public status page and a marketplace listing, custom service level agreements offered to enterprise customers. Nothing published on data residency, region choice, private cloud or customer hosted options, which matters for a vendor selling simultaneously into United Kingdom, European Union and United States insurers whose supervisors take different positions on where policy and quote data may sit.

Swallow publishes what it costs. Entry is 2,500 pounds a month, scaling on quotes processed and gross written premium, with 100,000 quotes included per plan. Insurers under 10 million pounds of premium get a published discount. The filings line publishes its own method. Raters are priced per state and per product, with a lower rate on each additional state. Filing assistance is priced per filing, with a lighter fee for each additional state and reduced fees for revisions. Monitoring and market analytics run on annual subscriptions tiered by coverage scope. The one soft spot is the up to fifty percent cheaper comparison, which names no baseline.

Coverage is broad for a company this size. Named institution types are carriers, reinsurers, managing general agents and program managers, along with captives, risk retention groups, brokers and embedded distribution partners. Geographies covered are the United Kingdom, the European Union and the United States, with state level awareness for filings. Lines are enumerated at length.

Personal lines run through private passenger auto, homeowners, dwelling fire, renters, condo, mobile home, umbrella, pet, motorcycle, recreational vehicle, boat, powersports and warranty. Commercial lines cover workers compensation, commercial auto, business owners policy, commercial property, general liability, professional liability, directors and officers, cyber, inland marine, equipment breakdown, surety and management liability.

Specialty lines include medical malpractice, trucking, ocean marine, aviation, equine and agricultural, and energy and environmental. The platform also states applicability to credit, mortgage, financing, leasing and membership pricing, though no customer outside insurance is named.