You sign the September 1 certification. We measure the model behind it.
Connecticut adopted the NAIC model AI bulletin as Bulletin MC-25, and domestic insurers certify annually, on or before September 1. That certification is your company's own statement. What we add underneath it is a dated record of how your model actually behaves.
Where a measurement fits.
MC-25 follows the NAIC model text, and you read it more closely than we do. The piece we work on is narrow. When someone asks how the model itself behaves, most files have adjectives where they need numbers.
The honest note, up front: nothing in MC-25 requires independent testing. The certification is your own statement, and we will never claim a regulation requires this measurement. A signature is easier to stand behind with a dated, independent record underneath it.
A signed, dated screening measurement.
We run your real question types against a model your company hosts and read the model's internal signals while it answers. The report shows where it fabricates, how often, on which topics, and names the flagged answers so your team can check each one before anyone signs. Published performance: 0.852 AUC on models it had never seen (fabricated-entity discrimination, length-controlled, leave-one-model-out), in the lab's method paper (DOI: 10.5281/zenodo.21365654).
Because MC-25 adopts the NAIC model text, the provision-by-provision crosswalk on this site maps the report to the same language, including what stays honestly outside scope. What we can measure is a model your company hosts or controls whose deployment exposes prompt-side token probabilities, such as vLLM-class private-cloud deployments and fine-tunes. Private Azure OpenAI exposes probabilities on output tokens only, so it takes behavioral testing instead. Vendor-attested SaaS AI is out of scope, and we say so. The test never touches policyholder data.
Still piloting? The pre-deployment Model Selection Study measures your candidate models before you commit to one.
Fixed fees, never success-based.
The LLM Validation Report is $9,500. Continuous Assurance is $6,500 per quarter for one model. The pre-deployment Model Selection Study runs $4,500–$7,500 by candidate count. The rest is on the pricing page, and no work begins without a signed engagement letter.
Getting this into the file before September 1.
Delivery is five business days from materials-complete, meaning your question types received and your endpoint reachable. Write the lab with your model setup and the certification date you are working toward, and you will get a written plan back the same business day.
Email the lab