Predictika · auto and home insurance quoting

An AI agent's insurance discounts, checked before the quote is released

An AI sales agent assembles a list of discounts for an auto and home insurance quote and states the total. Before the quote is released, the agent sends it to Predictika, which checks every discount against its conditions, the exclusions, the stated total and the stacking cap, and returns a verdict, every reason and the figures it computed.

Scope: Auto and home insurance discount eligibility and stacking: ten discounts, their conditions, exclusions, the stated total and the stacking cap. See everything this validator covers

This validator models discount eligibility and stacking on auto and home insurance quotes, not just the scenarios shown below.

  • Ten discounts — bundling, safe driver, anti-theft, good student, paperless, autopay, defensive driving, multi-car, 10-year loyalty and new customer, each with the percentage the quote carries for it.
  • Conditions — bundling needs both an auto and a home policy; safe driver needs safe-driver status and no at-fault accident in 3 years; anti-theft needs a device; good student needs the status and a student under 25; paperless and autopay need those billing options on; defensive driving needs a valid course; multi-car needs two or more vehicles; loyalty needs 10 or more years of continuous coverage; new customer needs a customer new to the carrier.
  • Exclusions, total and cap — the new-customer and loyalty discounts cannot be combined; the stated total must equal the sum of the proposed discounts; a total above the maximum stack needs an underwriter's approval.

Any discount quote that touches these rules is in scope — not only the scenarios pictured here. Not covered: premiums and base rates, other discounts and state filings. The percentages, the 30% maximum stack and the thresholds (under 25, 10 years, two vehicles) are typical values chosen for this model, not any specific insurer's program.

The AI sales agentAssembles the discounts, states the total and says the quote is ready to release.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

7 scenarios in 3 groups · each group runs from the agent’s first attempt to the outcome · click one to switch

What the agent proposes

The AI sales agent says

Illustrative agent statement. The validator receives the structured request below.

Proposing agent: AI sales agent (scripted)

Before sending, the agent must call Recorded run

Predictika validator


      

        
        
        
      

      
Waiting for the call.

      
    

  

  
  

    
Where rules-based systems get it wrong
Click to expand
    
Our assessment of what a typical rules-based CPQ (product rules, price rules and validation rules, as in Salesforce CPQ) does with the same cases. It is a judgment about effort and reliability, not a claim that nothing can be built.

    
What the validator caughtTypical rules-based quoting systemPredictika
A discount claimed without meeting its conditionOnly if someone is vigilant
Each condition is its own validation rule on its own screen; they work one by one, but each needs a rule and a test of its own.		All ten conditions are checked on every call, whatever order the facts arrive in.
Two discounts that cannot be combinedOnly if someone is vigilant
Exclusions are often a separate matrix that has to be kept in step with the eligibility rules.		The clash is one rule in the same model and is reported with the other findings.
A stated total that must equal the sum of the discountsOnly with workarounds
The total is typed or computed in another step, so a mismatch is found late, if at all.		The sum is computed on every call and compared with the stated total.
A total over the stacking cap that needs a person to sign offOnly with workarounds
A cap check usually passes or fails; sending the quote to an underwriter is a separate workflow step.		Returned as an approval finding, distinct from the blocking findings.
Every problem returned in one answerOnly with workarounds
Rules tend to fire one at a time as each step is completed, so one error is fixed and the next appears.		All findings come back together, whatever order the choices arrived in.

  

  

    
What this validator catches: findings from its test cases
Click to expand
    
Every rule is checked on every call, in any order the choices arrive; one problem never hides another. Blocking findings stop the discount quote; approval findings need a person to sign off; everything else is computed.

    
1The bundling discount needs both an auto policy and a home policy.
2The safe-driver discount needs safe-driver status and no at-fault accidents in the last 3 years.
3The anti-theft discount needs an anti-theft device on the vehicle.
4The good-student discount needs good-student status and a student under 25.
5The paperless discount needs paperless billing to be on.
6The autopay discount needs autopay to be on.
7The defensive-driving discount needs a valid course on file.
8The multi-car discount needs at least 2 vehicles; the policy has 1.
9The 10-year loyalty discount needs 10 or more years of continuous coverage; the customer has 9.
10The new-customer discount and the 10-year loyalty discount cannot be combined.
11The new-customer discount needs a customer who is new to the carrier.
12The stated total of 9.0% does not match the 5.0% that the proposed discounts add up to.
13The total discount of 31.0% is above the 30.0% maximum stack; an underwriter must approve it.

  

  

    
Scenarios and results
Click to expand
    
The recorded result of the validator for each scenario on this page: the request the AI agent sent, the verdict, and every finding, in the order they appear above.

    
1. A long-time customer, auto policy only: Eight discounts adding up to 44%

    
The agent's request. Please issue these discounts: bundling (10%), safe-driver (8%), anti-theft (5%), paperless (3%), autopay (2%), multi-car (4%), 10-year loyalty (7%), new-customer (5%), for a total of 44%. It is ready to release.

    
Verdict: Blocked.

    
    
      
  • Bundling needs auto and home. The bundling discount needs both an auto policy and a home policy.
    • 
      
  • New-customer and loyalty discounts clash. The new-customer discount and the 10-year loyalty discount cannot be combined.
    • 
      
  • Customer is not new. The new-customer discount needs a customer who is new to the carrier.
    • 
      
  • Total above the stacking cap. The total discount of 44.0% is above the 30.0% maximum stack; an underwriter must approve it.
    • 
    

    
2. A long-time customer, auto policy only: The agent resubmits after fixing the errors

    
The agent's request. Please issue these discounts: bundling (10%), safe-driver (8%), anti-theft (5%), paperless (3%), autopay (2%), multi-car (4%), 10-year loyalty (7%), for a total of 44%. It is ready to release.

    
Verdict: Blocked.

    
    
      
  • Bundling needs auto and home. The bundling discount needs both an auto policy and a home policy.
    • 
      
  • Stated total does not match. The stated total of 44.0% does not match the 39.0% that the proposed discounts add up to.
    • 
      
  • Total above the stacking cap. The total discount of 44.0% is above the 30.0% maximum stack; an underwriter must approve it.
    • 
    

    
3. A long-time customer, auto policy only: The agent resubmits after fixing the errors

    
The agent's request. Please issue these discounts: safe-driver (8%), anti-theft (5%), paperless (3%), autopay (2%), multi-car (4%), 10-year loyalty (7%), for a total of 29%. It is ready to release.

    
Verdict: Allowed.

    
    
      
  • Discount summary. The proposed discounts add up to 29.0%; the maximum stack is 30.0%.
    • 
    

    
4. Bundling without a home policy: Five discounts, including bundling, on an auto-only policy

    
The agent's request. Please issue these discounts: bundling (10%), paperless (3%), autopay (2%), defensive-driving (4%), multi-car (4%), for a total of 25%. The maximum stack is 30%. It is ready to release.

    
Verdict: Blocked.

    
    
      
  • Bundling needs auto and home. The bundling discount needs both an auto policy and a home policy.
    • 
      
  • Stated total does not match. The stated total of 25.0% does not match the 23.0% that the proposed discounts add up to.
    • 
    

    
5. Bundling without a home policy: The agent resubmits after fixing the errors

    
The agent's request. Please issue these discounts: paperless (3%), autopay (2%), defensive-driving (4%), multi-car (4%), for a total of 13%. The maximum stack is 30%. It is ready to release.

    
Verdict: Allowed.

    
    
      
  • Discount summary. The proposed discounts add up to 13.0%; the maximum stack is 30.0%.
    • 
    

    
6. A total just over the stacking cap: Four discounts adding up to 31%

    
The agent's request. Please issue these discounts: bundling (10%), safe-driver (8%), good-student (6%), 10-year loyalty (7%), for a total of 31%. It is ready to release.

    
Verdict: Needs approval.

    
    
      
  • Total above the stacking cap. The total discount of 31.0% is above the 30.0% maximum stack; an underwriter must approve it.
    • 
      
  • Discount summary. The proposed discounts add up to 31.0%; the maximum stack is 30.0%.
    • 
    

    
7. A total just over the stacking cap: The agent resubmits after fixing the errors

    
The agent's request. Please issue these discounts: bundling (10%), safe-driver (8%), 10-year loyalty (7%), for a total of 25%. It is ready to release.

    
Verdict: Allowed.

    
    
      
  • Discount summary. The proposed discounts add up to 25.0%; the maximum stack is 30.0%.
    • 
    

  

  

    
Narration transcript
Click to expand
    
The text of the narrated walkthrough, step by step.

    
    
      
  1. Welcome to a demo of Predictika's Validator. An AI sales agent assembles a list of discounts for an auto and home insurance quote and states the total. Before the quote is released, the agent sends it to Predictika, which checks every discount against its conditions, the exclusions, the stated total and the stacking cap, and returns a verdict, every reason and the figures it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please issue these discounts: bundling (10%), safe-driver (8%), anti-theft (5%), paperless (3%), autopay (2%), multi-car (4%), 10-year loyalty (7%), new-customer (5%), for a total of 44%. It is ready to release.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with three findings and one approval, all in one answer. The bundling discount needs both an auto policy and a home policy. The new-customer discount and the 10-year loyalty discount cannot be combined. The new-customer discount needs a customer who is new to the carrier.
    6. 
      
  6. The validator returned three blocking findings and one approval in one answer. A check that reads one field at a time would pass most of this; it shows only when the choices in the discount quote are checked against each other.
    7. 
      
  7. What this shows: for a total over the stacking cap that needs a person to sign off, a rules-based quoting system can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the discounts proposed to bundling, safe-driver, anti-theft, paperless, autopay, multi-car, 10-year loyalty.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Still blocked, with two findings and one approval, all in one answer. The bundling discount needs both an auto policy and a home policy. The stated total of 44.0% does not match the 39.0% that the proposed discounts add up to.
    12. 
      
  12. Two findings cleared. Still blocking: bundling needs auto and home. Newly found: stated total does not match. One approval also remains for a person: total above the stacking cap.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It changes the discounts proposed to safe-driver, anti-theft, paperless, autopay, multi-car, 10-year loyalty and the total stated by the agent to 29%.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. All three earlier findings are cleared, and the validator's figure matches the discount quote.
    18. 
      
  18. Four problems caught in one call, fixed in two passes, and a clean discount quote. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.
    19.