InsuranceExam-Ready Toolkit5 · Workflow map

Part 5 of The Exam-Ready Toolkit

Where the guessing is allowed to happen

Workflow map

Version 0.1Updated Aug 24, 2026Anchored to NAIC AI Risk Evaluation Supplement draft 4.0
On this page

    The Supplement classifies AI by autonomy: support, augment, automate. Architecture classifies by certainty: deterministic, probabilistic, human. Governance is where the two meet. A probabilistic component inside a deterministic control structure is governable. A probabilistic component that is the control structure is a lawsuit with a timestamp.


    The rule

    Three territories, one design pattern.

    Deterministic: same inputs, same output, every time, provably. This is about reproducibility, not about whether machine learning is involved. A trained model with fixed weights and a pinned version is deterministic at inference; a language model sampled at non-zero temperature is not. A GLM or gradient-boosted rating model that is filed, versioned, and locked belongs here. Anything that has to be filed, reconciled, reproduced for an examiner, or that moves money. Rating, binding, issuance, pro-rata math, billing, reserves, payouts, bordereaux. A rating algorithm that gives different premiums for identical inputs isn’t innovative. It’s unfileable.

    Probabilistic: pattern work where 95% right at a tenth of the cost is a good trade because the failure mode is a correctable inefficiency. Extraction from unstructured documents, classification, triage, drafting, summarisation, anomaly flagging, similarity search.

    Human: accountability joints. Adverse decisions, exceptions, anything irreversible, anything where the examiner will ask “who decided.”

    The pattern that survives every regulatory regime: the model proposes, the rule disposes, the human owns the exceptions. The probabilistic component emits a score or a structured suggestion. A deterministic rule consumes it and routes. A human handles what the rule sends them. The score, the rule version, and the human action are all logged (Control 4).

    The most common failure isn’t a bad model. It’s a model whose output nothing deterministic receives. That is Nolte’s architectural reading of what has been alleged in the nH Predict litigation, which is ongoing: not that a model was wrong, but that, as alleged, no mechanism existed to catch it being wrong.

    The second most common failure is spending probability where the business needed certainty. A six-figure monthly inference bill on quote-bind-issue is what that looks like in production.


    The map

    For each workflow: the steps, the placement, the tier from Part 3, the Exhibit A operational area, and the failure mode if placement is wrong.

    Submission and intake

    StepPlacementWhy
    Receive submission (email, portal, ACORD form, broker PDF)Deterministic ingestIdempotent, logged, no interpretation
    Extract fields from unstructured docsProbabilisticPattern work; wrong field is caught downstream
    Validate extracted fields against schema and business rulesDeterministicRequired fields, ranges, state eligibility
    Human confirmation of low-confidence extractionsHumanConfidence threshold set in config (Control 3)
    Appetite check and routingDeterministic rule on the extracted, validated dataThe rule can consume a model score, but the rule decides

    Tier: 1 (extraction) rising to 2 if the appetite routing consumes a model score that could decline a risk. Exhibit A: Underwriting/Eligibility, Producer Services. Failure if wrong: an LLM “deciding” a risk is out of appetite. That’s a declination with no rule to point at.

    Quoting and rating

    StepPlacementWhy
    Rating engine (factors, relativities, tiering, GLM outputs)Deterministic, versioned, lockedFiled. Must reproduce. Same inputs, same premium, forever.
    Quote assembly, forms selection, disclosuresDeterministicForm applicability is rule-based per state
    Quote explanation or summary for the customer or brokerProbabilistic, on top of deterministic outputsDrafting from structured data; wrong wording is embarrassing, not unfileable
    Referral decision (auto-quote vs refer to underwriter)Deterministic rule, may consume a model scoreThe score is an input; the rule is the decision
    Underwriter judgement on referralsHumanAccountability joint

    Tier: 3 for the rating engine (automate, material financial impact, consumer impact). 1 for the summary. 2 for a model-scored referral rule. Exhibit A: Premium Quotes and Discounts, Ratemaking/Rate Classification. Failure if wrong: any probabilistic component between the validated inputs and the premium. It doesn’t matter how accurate it is; it cannot be filed.

    Binding and issuance

    StepPlacementWhy
    Bind eligibility checks (payment, signatures, KYC, state rules)DeterministicCompliance gate
    Policy number, effective dates, document generationDeterministicReproducible artifacts
    Carrier notification, bordereaux row creation at bindDeterministicThis is the data-model decision: structure it now, not at reporting time
    Fraud or identity anomaly flag at bindProbabilistic score → deterministic hold rule → human reviewModel proposes, rule disposes

    Tier: 3 for bind itself. 2 for the fraud score if it can hold a bind. Exhibit A: Underwriting/Eligibility, Fraud. Failure if wrong: a probabilistic “should we bind” step. Binding is a contract formation event. It’s deterministic or it’s not binding.

    Endorsements and mid-term changes

    StepPlacementWhy
    Interpret the change request (email, call note, portal free text)Probabilistic extraction to a structured change objectPattern work; validated next
    Validate the change against policy, product, and state rulesDeterministicWhich endorsements are allowed, when, with what forms
    Re-rate and pro-rata premium calculationDeterministicMoney math. Never probabilistic.
    Adverse changes (coverage reduction, cancellation, non-renewal notice)Deterministic rule → human approvalAdverse consumer outcome; human owns it
    Endorsement document generationDeterministicReproducible

    Tier: 1 for interpretation. 3 for re-rate. 3 for adverse changes. Exhibit A: Underwriting/Eligibility, Customer Service. Failure if wrong: the extraction silently changing a limit and the deterministic layer trusting it. Validation must be against the policy of record, not the extracted object.

    Certificates of insurance

    StepPlacementWhy
    Intake of COI request (broker email, portal, holder form)Probabilistic extraction of holder, requirements, datesPattern work
    Match requested coverages against policy of recordDeterministicThe certificate must reflect the policy exactly
    Flag requests that exceed the policy (additional insured not on policy, limits above bound)Deterministic rule → human or brokerExceptions are where errors and E&O exposure live
    Generate certificateDeterministic templating from policy dataReproducible, auditable
    Renewal-driven reissueDeterministic, scheduled

    Tier: 1 for intake. 2 for the whole flow because a wrong certificate is a misrepresentation to a third party. Exhibit A: Customer Service, Producer Services. Failure if wrong: an LLM drafting the certificate. The certificate is a statement of fact about a contract; it comes from the contract, not from a model’s reading of an email.

    Claims

    StepPlacementWhy
    FNOL intake (call, form, photos, documents)Probabilistic extraction and classificationPattern work; the biggest legitimate win in the stack
    Coverage determination (is this loss covered under this policy, on this date, for this peril)Deterministic against policy of recordNever a model. This is the contract.
    Triage and routing (severity, complexity, fraud signal, assignment)Probabilistic score → deterministic routing ruleThe rule can be tuned; the score is logged
    Reserve settingDeterministic or actuarial method, versioned; model may proposeFinancial statement impact (Exhibit C ref 11)
    Adjudication: approve within authorityDeterministic rule on validated facts, possibly model-assistedAutomate only below a written authority limit
    Adjudication: deny, partial, disputeHuman, model may summariseAdverse consumer outcome; unfair claims settlement exposure
    Payout calculation and paymentDeterministicMoney
    Subrogation and salvage identificationProbabilistic flag → humanRecovery opportunity; wrong flag costs little
    Customer communication draftingProbabilistic draft → deterministic compliance check (required language, timelines) → human send for adverseTimelines are statutory per state

    Tier: 3 for coverage determination, adjudication, reserves, payout. 2 for triage. 1 for FNOL extraction and comms drafting. Exhibit A: Claims/Adjudication. Failure if wrong: a model producing the coverage determination or the denial. This is exactly where the industry has already been sued, and exactly where the Supplement’s Exhibit C ref 12 points (unfair claims settlement laws).

    Renewals, cancellations, non-renewals

    StepPlacementWhy
    Renewal eligibility and re-rateDeterministicFiled
    Retention risk scoringProbabilisticMarketing-adjacent; wrong score costs a call
    Non-renewal or conditional renewal decisionDeterministic rule → humanAdverse; state notice periods are statutory
    Notice generation and timingDeterministicStatutory timelines

    Tier: 3 for non-renewal. 1 for retention scoring. Exhibit A: Underwriting, Customer Service, Marketing.

    Bordereaux and carrier reporting

    StepPlacementWhy
    Row creationDeterministic, at bind and at every transactionStructure at the source
    Reconciliation against premium and claims ledgersDeterministicMust balance
    Anomaly detection across the file before submissionProbabilistic flag → human reviewCatches what reconciliation doesn’t
    Submission to carrierDeterministic, versioned, receiptedEvidence of what was sent

    Tier: 3 (financial reporting impact). Exhibit A: Reinsurance, Legal/Compliance, Reserves. Failure if wrong: generating bordereaux from unstructured data with a model. Bordereaux is a data-model decision made at bind time, not a reporting feature bolted on after.

    Billing and payments

    Entirely deterministic except: dunning message drafting (probabilistic, compliance-checked) and payment fraud scoring (probabilistic → rule → human). Tier: 3 for money movement. Exhibit A: Other.

    Customer service and producer support

    StepPlacementWhy
    Answering policy questions from documentsProbabilistic, grounded in policy of record, with citationsWrong answer costs a follow-up
    Anything that sounds like a coverage confirmationDeterministic lookup, or human”You’re covered” is a decision
    Complaint intake and taggingProbabilistic classification → deterministic tag to the AI system involved (checklist 3m)

    Tier: 1, rising to 2 the moment the bot confirms coverage. Exhibit A: Customer Service, Producer Services.


    Reading the map

    Count the cells. Probabilistic placements cluster at three points: intake (extraction, classification), scoring (triage, fraud, retention), and drafting (summaries, communications). Everything between those points and the money is deterministic. Everything adverse has a human at the end.

    That is not a limitation on AI. It’s where the wins are. FNOL extraction, submission intake, and endorsement interpretation are where operators lose the most time, and they’re all Tier 1. The expensive part of the stack was never the part that needed a model.

    Three tests for any proposed placement:

    1. Does the examiner need to reproduce this output? Deterministic.
    2. Does this output move money or form a contract? Deterministic.
    3. Is this output adverse to a consumer? Human at the end, deterministic rule before. If none apply, probabilistic is allowed, provided something deterministic receives the output and Control 4 logs it.

    What we build this way to protect

    In Nolte’s own delivery experience across a full MGA stack in all fifty states, the hard part was never the code. It was drawing this line and holding it under pressure from every vendor demo and every “just let the model handle it” suggestion. The rating engine was deterministic, versioned, locked. Quote-bind-issue was deterministic. Bordereaux was deterministic. The probabilistic tools lived at the intake edge where being wrong was cheap.

    That line is what the Supplement now asks you to draw in writing. Exhibit A’s autonomy column, Exhibit C’s risk classification, and Exhibit B Q1g’s autonomy-and-reversibility question are all the same question: where did you allow the guessing, and what catches it.


    What is verified

    What is Nolte's read

    • Every placement in the map: these are design positions from delivery experience, not regulatory text
    • The nH Predict characterisation is Nolte's architectural reading of allegations in ongoing litigation, not a finding of fact

    Self-assessment and orientation only. Nothing here is legal advice, a certification, or a compliance determination. Nolte is not affiliated with or endorsed by the NAIC. Exhibit references are paraphrased from Supplement draft 4.0; the wording that counts is the NAIC's. Corrections: j@nolte.io.