Predictika · Cyber insurance

An AI sales agent's cyber insurance quote, checked before it goes out

An AI sales agent prepares a Beazley Breach Response cyber quote for an applicant: the industry and revenue, the aggregate limit, the retention, the eCrime sublimit, the Reputation Loss add-on, the notified individuals, the waiting period, the applicant's security controls and the premium the quote states. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how those choices fit together and returns a verdict, every reason and the premium it computed.

Scope: One Breach Response cyber quote: limit, retention, eCrime sublimit, Reputation Loss, notified individuals, waiting period, controls and the stated premium. See everything this validator covers

This validator models Breach Response cyber quotes, not just the scenarios shown below.

  • Limit and retention — the aggregate limits offered, the largest limit and the smallest retention that each revenue band allows, and the retentions offered.
  • Coverage parts — the eCrime sublimit against its cap and the aggregate limit, Reputation Loss, which needs a larger limit, the number of notified individuals for breach response and the waiting period.
  • The applicant's controls — multi-factor authentication on remote access, offline backups, a prior cyber incident and a large healthcare record count, which go to an underwriter or stop the quote.
  • Premium — the premium computed from the limit, the industry, the revenue band and the retention credit, a referral above 100,000 dollars, and the stated premium against the one computed.

Any Breach Response quote that touches these rules is in scope — not only the scenarios pictured here. Not covered: endorsements other than Reputation Loss, taxes, fees and other Beazley products. Coverage names and the structure of the schedule follow Beazley's public Breach Response materials; the limit and retention bands, the controls and referral rules and every premium rate are modeled from comparable cyber products and are illustrative, not published by Beazley.

The AI sales agentBuilds the cyber quote from the conversation with the broker and says it is ready to send.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

6 scenarios in 2 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 CPQ gets 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 CPQPredictika
Limits that depend on the revenue bandOnly with care
The largest limit and the smallest retention both change with the applicant's revenue, so each band needs its own pair of rules, and a new band means new rules.		The band is looked up from the revenue and both bounds are checked in the same pass.
A coverage part that needs a larger limitOnly with workarounds
Reputation Loss is offered only above a limit and the eCrime sublimit must stay under both its cap and the aggregate limit, so each part needs a rule that reads another field.		Each part is checked against the aggregate limit chosen in the same call.
Controls that stop or refer a quoteOnly with workarounds
Missing multi-factor authentication stops the quote, while no offline backups, a prior incident or a large healthcare record count only refer it, and the rules for each fire at different times.		Every control is evaluated together, and the answer says which stop the quote and which go to an underwriter.
A stated premium that must match the computed oneOnly with workarounds
The premium depends on the limit, the industry, the revenue band, the retention credit and the add-ons, so a separate rule has to compute it and compare it with the premium the quote states.		The premium is computed inside the model and compared with the stated one.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the quote is saved or a step is completed, so a broker often fixes one error and meets the next.		All findings come back together, so the agent can fix them in a single pass.

  

  

    
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 quote; approval findings need a person to sign off; everything else is computed.

    
1An aggregate limit of 4,000 thousand dollars is not one of the limits offered (1, 2, 3, 5 or 10 million).
2An aggregate limit of 10,000 thousand dollars is above the 5,000 thousand maximum for this revenue band.
3A retention of 75 thousand dollars is not one of the retentions offered (10, 25, 50, 100 or 250 thousand).
4A retention of 10 thousand dollars is below the 25 thousand minimum for this revenue band.
5An eCrime sublimit of 251 thousand dollars is above the 250 thousand maximum or above the aggregate limit.
6Breach response for 1,001 thousand notified individuals is above the 1,000 thousand (1,000,000) maximum.
7A waiting period of 6 hours is below the 8-hour minimum.
8Reputation Loss is offered only with an aggregate limit of at least 2,000 thousand dollars.
9The applicant has no multi-factor authentication on remote access; this risk is not quoted.
10The applicant keeps no offline or immutable backups; an underwriter should review ransomware exposure.
11The applicant reports a cyber incident in the last three years; an underwriter should review it.
12A healthcare applicant holding 1,001 thousand records needs an underwriter review.
13A premium of 146,300 dollars is over the 100,000 dollar referral threshold.
14The quote states a premium of 11,599 dollars; the modeled rates give 11,600 dollars.

  

  

    
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. Healthcare applicant quote: A healthcare quote with a limit that is not offered, a retention below the minimum, an eCrime sublimit over its cap, too many notified individuals, a waiting period that is too short, no multi-factor authentication on remote access and a wrong premium, for an applicant with no offline backups and a large record count

    
The agent's request. Please quote Beazley Breach Response for a healthcare applicant with 60 million dollars of revenue and 1,500,000 records. Aggregate limit: 4,000,000 dollars; retention: 10,000 dollars; eCrime sublimit: 300,000 dollars. The Reputation Loss add-on is included. Breach response is for 1,200,000 notified individuals and the waiting period is 4 hours. Multi-factor authentication on remote access: no; offline backups: no. The quote states a premium of 20,000 dollars. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Aggregate limit not offered. An aggregate limit of 4,000 thousand dollars is not one of the limits offered (1, 2, 3, 5 or 10 million).
    • 
      
  • Retention below the minimum. A retention of 10 thousand dollars is below the 25 thousand minimum for this revenue band.
    • 
      
  • eCrime sublimit too high. An eCrime sublimit of 300 thousand dollars is above the 250 thousand maximum or above the aggregate limit.
    • 
      
  • Notified individuals above the maximum. Breach response for 1,200 thousand notified individuals is above the 1,000 thousand (1,000,000) maximum.
    • 
      
  • Waiting period too short. A waiting period of 4 hours is below the 8-hour minimum.
    • 
      
  • No multi-factor authentication on remote access. The applicant has no multi-factor authentication on remote access; this risk is not quoted.
    • 
      
  • Stated premium does not match. The quote states a premium of 20,000 dollars; the modeled rates give 39,480 dollars.
    • 
      
  • No offline backups. The applicant keeps no offline or immutable backups; an underwriter should review ransomware exposure.
    • 
      
  • Large healthcare record count. A healthcare applicant holding 1,500 thousand records needs an underwriter review.
    • 
    

    
2. Healthcare applicant quote: The broker resubmits after fixing the errors

    
The agent's request. Please quote Beazley Breach Response for a healthcare applicant with 60 million dollars of revenue and 1,500,000 records. Aggregate limit: 3,000,000 dollars; retention: 25,000 dollars; eCrime sublimit: 250,000 dollars. The Reputation Loss add-on is included. Breach response is for 1,200,000 notified individuals and the waiting period is 4 hours. Multi-factor authentication on remote access: yes; offline backups: no. The quote states a premium of 28,202 dollars. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Notified individuals above the maximum. Breach response for 1,200 thousand notified individuals is above the 1,000 thousand (1,000,000) maximum.
    • 
      
  • Waiting period too short. A waiting period of 4 hours is below the 8-hour minimum.
    • 
      
  • No offline backups. The applicant keeps no offline or immutable backups; an underwriter should review ransomware exposure.
    • 
      
  • Large healthcare record count. A healthcare applicant holding 1,500 thousand records needs an underwriter review.
    • 
    

    
3. Healthcare applicant quote: The broker resubmits after fixing the errors

    
The agent's request. Please quote Beazley Breach Response for a healthcare applicant with 60 million dollars of revenue and 800,000 records. Aggregate limit: 3,000,000 dollars; retention: 25,000 dollars; eCrime sublimit: 250,000 dollars. The Reputation Loss add-on is included. Breach response is for 1,000,000 notified individuals and the waiting period is 8 hours. Multi-factor authentication on remote access: yes; offline backups: yes. The quote states a premium of 28,202 dollars. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Quote summary. 3,000 thousand dollar aggregate limit, 25 thousand retention, modeled premium 28,202 dollars (maximum limit 5,000 thousand and minimum retention 25 thousand for this revenue band).
    • 
    

    
4. Retail applicant quote with Reputation Loss: A retail quote with the Reputation Loss add-on on a 1 million dollar limit, a retention below the minimum for the revenue band, an eCrime sublimit above the limit, a wrong premium and no offline backups

    
The agent's request. Please quote Beazley Breach Response for a retail applicant with 400 million dollars of revenue and 200,000 records. Aggregate limit: 1,000,000 dollars; retention: 50,000 dollars; eCrime sublimit: 1,200,000 dollars. The Reputation Loss add-on is included. Breach response is for 500,000 notified individuals and the waiting period is 8 hours. Multi-factor authentication on remote access: yes; offline backups: no. The quote states a premium of 5,000 dollars. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Retention below the minimum. A retention of 50 thousand dollars is below the 100 thousand minimum for this revenue band.
    • 
      
  • eCrime sublimit too high. An eCrime sublimit of 1,200 thousand dollars is above the 250 thousand maximum or above the aggregate limit.
    • 
      
  • Reputation Loss needs a larger limit. Reputation Loss is offered only with an aggregate limit of at least 2,000 thousand dollars.
    • 
      
  • Stated premium does not match. The quote states a premium of 5,000 dollars; the modeled rates give 11,148 dollars.
    • 
      
  • No offline backups. The applicant keeps no offline or immutable backups; an underwriter should review ransomware exposure.
    • 
    

    
5. Retail applicant quote with Reputation Loss: The broker resubmits after fixing the errors

    
The agent's request. Please quote Beazley Breach Response for a retail applicant with 400 million dollars of revenue and 200,000 records. Aggregate limit: 2,000,000 dollars; retention: 100,000 dollars; eCrime sublimit: 250,000 dollars. The Reputation Loss add-on is included. Breach response is for 500,000 notified individuals and the waiting period is 8 hours. Multi-factor authentication on remote access: yes; offline backups: no. The quote states a premium of 17,024 dollars. It is ready to send.

    
Verdict: Needs approval.

    
    
      
  • No offline backups. The applicant keeps no offline or immutable backups; an underwriter should review ransomware exposure.
    • 
      
  • Quote summary. 2,000 thousand dollar aggregate limit, 100 thousand retention, modeled premium 17,024 dollars (maximum limit 10,000 thousand and minimum retention 100 thousand for this revenue band).
    • 
    

    
6. Retail applicant quote with Reputation Loss: The broker resubmits after fixing the errors

    
The agent's request. Please quote Beazley Breach Response for a retail applicant with 400 million dollars of revenue and 200,000 records. Aggregate limit: 2,000,000 dollars; retention: 100,000 dollars; eCrime sublimit: 250,000 dollars. The Reputation Loss add-on is included. Breach response is for 500,000 notified individuals and the waiting period is 8 hours. Multi-factor authentication on remote access: yes; offline backups: yes. The quote states a premium of 17,024 dollars. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Quote summary. 2,000 thousand dollar aggregate limit, 100 thousand retention, modeled premium 17,024 dollars (maximum limit 10,000 thousand and minimum retention 100 thousand for this revenue band).
    • 
    

  

  

    
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 prepares a Beazley Breach Response cyber quote for an applicant: the industry and revenue, the aggregate limit, the retention, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how those choices fit together and returns a verdict, every reason and the premium it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote Beazley Breach Response for a healthcare applicant with 60 million dollars of revenue and 1,500,000 records. Aggregate limit: 4,000,000 dollars; retention: 10,000 dollars; eCrime sublimit: 300,000 dollars. The rest of the request is on screen. The quote states a premium of 20,000 dollars.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with seven findings and two approvals, all in one answer. An aggregate limit of 4,000 thousand dollars is not one of the limits offered (1, 2, 3, 5 or 10 million). A retention of 10 thousand dollars is below the 25 thousand minimum for this revenue band. An eCrime sublimit of 300 thousand dollars is above the 250 thousand maximum or above the aggregate limit. The other four are on screen.
    6. 
      
  6. The validator returned seven blocking findings and two approvals 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 quote are checked against each other.
    7. 
      
  7. What this shows: for a stated premium that must match the computed one, a rules-based CPQ can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the aggregate limit to 3,000,000 dollars, the retention to 25,000 dollars, the eCrime sublimit to 250,000 dollars, the multi-factor authentication on remote access to yes and the premium stated on the quote to 28,202 dollars.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Still blocked, with two findings and two approvals, all in one answer. Breach response for 1,200 thousand notified individuals is above the 1,000 thousand (1,000,000) maximum. A waiting period of 4 hours is below the 8-hour minimum.
    12. 
      
  12. Five findings cleared. Still blocking: notified individuals above the maximum and waiting period too short. Two approvals also remain for a person: no offline backups and large healthcare record count.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It changes the records held to 800,000, the notified individuals to 1,000,000, the waiting period to 8 hours and the offline backups to yes.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. All four earlier findings are cleared.
    18. 
      
  18. Nine problems caught in one call, fixed in two passes, and a clean 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.