Indianapolis – September 18, 2026 -- Kristen Nunery, CEO of illumend and founder of myCOI, is urging risk and compliance leaders to demand AI systems that do more than ground decisions in an organization's insurance requirements, arguing that grounding alone leaves compliance determinations vulnerable to costly errors. Nunery outlines her position in a new article, "Grounding Is Half the Answer in COI Compliance," aimed at organizations in construction, commercial real estate, retail, franchising and government that rely on vendors, contractors, tenants and franchisees to maintain contractually required insurance coverage.
Grounded AI sets the standard but cannot guarantee an accurate compliance decision
Grounded AI anchors its outputs to an organization's own contracts, leases, insurance requirements, vendor classifications and approved exceptions rather than general knowledge from model training. Nunery said this shifts the question a system answers from whether a vendor appears "adequately insured" -- a term with no universal definition -- to whether submitted evidence satisfies the specific requirements governing a given third-party relationship.
"Grounding gives the AI the standard instead of asking it to invent one," Nunery said. "A system that cannot tell you where its definition of compliant came from does not really have one."
Cross-document reasoning failures can produce wrong conclusions despite accurate readings
Nunery cited a scenario in which a certificate shows additional insured coverage while an attached endorsement covers only ongoing operations. If the governing agreement requires completed operations coverage, approving the vendor on the certificate alone could leave the organization exposed to a claim filed after work ends -- even though every individual document was read correctly.
"COI compliance is a document-relationship problem before it is a document-reading problem," Nunery said. "Cross-document reasoning tells the system what the complete evidence means, while grounding tells it which requirements that evidence must satisfy. Remove either one, and the resulting determination is a guess wearing a confident tone."
Auditability and human sign-off are positioned as non-negotiable controls
Nunery argues every AI-supported compliance determination should identify which requirement applied, which documents were reviewed and why the system reached its conclusion, so it can withstand an audit, dispute or claim. A confidence score or pass-fail label, she said, does not establish that a system applied the correct agreement or evaluated the complete body of evidence. A person should approve a determination before a third party is declared compliant, particularly when evidence is incomplete or an exception applies.
Seven vendor questions target requirements, evidence and error measurement
Nunery recommends buyers ask vendors who defines "compliant" in the system, whether requirements are configured before scale-up, whether certificates are connected to contracts and endorsements, whether missing evidence can be distinguished from deficient coverage, and whether a human approves determinations. She also urges organizations to track both false passes, which leave risk uncovered, and false failures, which generate unnecessary work chasing compliant vendors. "A system that reports only one of those numbers is telling you half of what happened," Nunery said.
illumend's platform ties requirements, evidence, reasoning and approval into one workflow
illumend, backed by myCOI's data foundation, configures an organization's insurance requirements during onboarding based on its contracts and leases, then evaluates submitted evidence, reviews endorsements, flags deficiencies, manages renewals and documents reasoning behind each determination, with human sign-off before a third party is notified. myCOI, founded by Nunery more than 16 years ago, has reviewed 45 million documents, managed 1.2 million agreements and identified 2 million coverage gaps before they became claims.