Predictika · Specialty insurance quotes

An AI sales agent's general liability quote, checked before it goes out

An AI sales agent prepares a quote for a wholesale broker on a casualty (general liability) policy: product, insured segment, revenue or project value, limit, retention or excess attachment and the premium. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the Bowhead Specialty Casualty product line as described in its public registration statement and returns a verdict, every reason, a line-item quote and the figures it computed.

Scope: the six products of the Bowhead Specialty Casualty division (Primary and Excess; Projects, Practice and Other): segments, wholesale-only distribution, limits up to $15 million, retention, excess attachment and premium. See everything this validator covers

This validator models general liability quotes for the Casualty division products below, not just the scenarios shown.

  • Products — Excess Projects (excess over large single construction projects), Excess Practice (annually renewable excess GL, products and auto liability for middle market contractors, typically $100 million to $1 billion of revenue), Excess Other (real estate, hospitality, public entity and manufacturing), Primary Projects (wrap-up GL on single projects), Primary Practice (GL for general contractors and subcontractors under $100 million of revenue) and Primary Other (manufacturers and distributors under $200 million of revenue). All six are real Bowhead product names.
  • Rules — wholesale-broker distribution only, a maximum limit of $15 million, about $5 million generally retained per policy (larger limits go to a person for the reinsurance structure), wrap-up only on Primary Projects, excess over underlying limits, segment fit and revenue bands.
  • Premium — exposure times a rate, a segment factor, a limit factor, a retention credit or excess layer factor and a minimum premium, added into a line-item quote and compared with the premium the agent states. The rates, factors, retentions and thresholds are typical values for this kind of policy, not Bowhead's rates.

Any general liability quote for these six products that touches these rules is in scope — not only the scenarios pictured here. Not covered: the Professional Liability and Healthcare divisions, surplus lines taxes and fees, state filings, endorsement wording, and specific forms. Nothing here is a Bowhead rate, appetite statement or underwriting guideline.

The AI sales agentBuilds the casualty quote for a wholesale broker and says it is ready.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

6 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 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
Stated premiums that must equal what the rating givesOnly with care
A stated figure is a free-entry field; a rule can compare it with a computed premium, but someone has to keep the comparison in step with every rating step.
The stated premium travels with the quote and is compared with the computed premium on every call.
Excess layers whose factor depends on the underlying limits, plus a minimum premiumOnly with care
The layer factor, limit factor and minimum premium are chained price rules; the order and the rounding at each step are easy to get wrong.
Every step is computed in the same pass and the line items reconcile to the total.
Which products fit which segment, revenue band and endorsementOnly with workarounds
Six products with different eligibility, wrap-up on one product only, and soft revenue bands that need approval rather than a block: a lot of cross-field rules that interact.
Product, segment, revenue, limit, retention and endorsement are in one request and checked together.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the quote is saved or a step is completed, so a rep often fixes one error and meets the next.
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 quote; approval findings need a person to sign off; everything else is computed.

1Casualty products are distributed only through wholesale brokers; a retail agent submission cannot be quoted.
2A limit of $15 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.
3A limit of $16 million is above the $15 million maximum limit on a policy.
4A limit of $16 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.
5A limit of $4 million is not one of the limits modeled (1, 2, 3, 5, 10 or $15 million).
6A limit of $10 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.
7The insured segment is not one this Casualty product is described for (projects and practice products are for construction; Primary Other is for manufacturers and distributors; Excess Other is for real estate, hospitality, public entity and manufacturing).
8The insured revenue of $100 million is outside the revenue band the product is described for (Primary Practice under 100 million, Excess Practice 100 million to 1 billion, Primary Other under 200 million); an underwriter should confirm.
9The insured revenue is $0 million; at least 1 million is needed.
10A retention of $10,000 on an insured with $25 million of revenue should be confirmed by an underwriter (modeled).
11A retention of $10,000 on an insured with $49 million of revenue should be confirmed by an underwriter (modeled).
12An insured with $50 million of revenue needs a retention of at least $25,000 (modeled).
13A retention of $30,000 is not one of the retentions modeled (10, 25, 50, 100 or $250,000).
14A primary policy cannot sit over underlying limits; the quote lists $5 million of underlying limits.
15Wrap-up general liability is written on the Primary Projects product; it is requested on another product.
16The insured revenue of $200 million is outside the revenue band the product is described for (Primary Practice under 100 million, Excess Practice 100 million to 1 billion, Primary Other under 200 million); an underwriter should confirm.
17The projects products are for large, complex projects; a project value of $24 million is below the $25 million minimum modeled.
18The project includes for-sale residential units; an underwriter should review the project.
19The insured revenue of $99 million is outside the revenue band the product is described for (Primary Practice under 100 million, Excess Practice 100 million to 1 billion, Primary Other under 200 million); an underwriter should confirm.
20The insured revenue of $1001 million is outside the revenue band the product is described for (Primary Practice under 100 million, Excess Practice 100 million to 1 billion, Primary Other under 200 million); an underwriter should confirm.
21An excess policy sits over underlying limits of at least $1 million (modeled); the quote lists less than that.
22The policy attaches above $10 million of underlying limits; an underwriter should confirm the layer.
23The projects products are for large, complex projects; a project value of $10 million is below the $25 million minimum modeled.
24The quote states a premium of $417,950; the rating gives $418,950.
25The quote states a premium of $418,951; the rating gives $418,950.
26An estimated premium of $1,071,000 is at or above $1,000,000 and needs a person to confirm.
27An estimated premium of $1,652,400 is at or above $1,000,000 and needs a person to confirm.
28A limit of $20 million is above the $15 million maximum limit on a policy.
29A primary policy cannot sit over underlying limits; the quote lists $2 million of underlying limits.
30A limit of $20 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.
31The insured revenue of $150 million is outside the revenue band the product is described for (Primary Practice under 100 million, Excess Practice 100 million to 1 billion, Primary Other under 200 million); an underwriter should confirm.
32The policy attaches above $15 million of underlying limits; an underwriter should confirm the layer.

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. Channel, limit and retention: General contractor submitted through a retail agent

The agent's request. Please quote a Primary Practice policy with a $10 million limit for a construction contractor, submitted through a retail agent with $80 million of annual revenue, with a $10,000 retention. The quote states an estimated premium of $772,800. It is ready to send.

Verdict: Blocked.

  • Retail submission. Casualty products are distributed only through wholesale brokers; a retail agent submission cannot be quoted.
  • Retention too low for the revenue. An insured with $80 million of revenue needs a retention of at least $25,000 (modeled).
  • Limit above the usual net retention. A limit of $10 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.

2. Channel, limit and retention: The agent resubmits after fixing the errors

The agent's request. Please quote a Primary Practice policy with a $10 million limit for a construction contractor, submitted through a wholesale broker with $80 million of annual revenue, with a $50,000 retention. The quote states an estimated premium of $695,520. It is ready to send.

Verdict: Needs approval.

  • Limit above the usual net retention. A limit of $10 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.
  • Quote summary. Estimated premium of $695,520 for a $10 million limit, with a minimum premium of $25,000.

3. Channel, limit and retention: The agent resubmits after fixing the errors

The agent's request. Please quote a Primary Practice policy with a $5 million limit for a construction contractor, submitted through a wholesale broker with $80 million of annual revenue, with a $50,000 retention. The quote states an estimated premium of $529,200. It is ready to send.

Verdict: Allowed.

  • Quote summary. Estimated premium of $529,200 for a $5 million limit, with a minimum premium of $25,000.

4. Wrap-up project: Excess Projects policy with wrap-up and no underlying limits

The agent's request. Please quote an Excess Projects policy with a $10 million limit for a construction contractor, submitted through a wholesale broker with $900 million of annual revenue and a single project valued at $180 million, written as wrap-up general liability. The quote states an estimated premium of $173,880. It is ready to send.

Verdict: Blocked.

  • Wrap-up only on Primary Projects. Wrap-up general liability is written on the Primary Projects product; it is requested on another product.
  • Excess without underlying limits. An excess policy sits over underlying limits of at least $1 million (modeled); the quote lists less than that.
  • Limit above the usual net retention. A limit of $10 million is above the roughly $5 million Bowhead generally retains per policy; the reinsurance structure should be confirmed.

5. Wrap-up project: The agent resubmits after fixing the errors

The agent's request. Please quote a Primary Projects policy with a $5 million limit for a construction contractor, submitted through a wholesale broker with $900 million of annual revenue and a single project valued at $180 million, with a $100,000 retention, written as wrap-up general liability. The quote states an estimated premium of $481,950. It is ready to send.

Verdict: Allowed.

  • Quote summary. Estimated premium of $481,950 for a $5 million limit, with a minimum premium of $25,000.

6. Segment fit: Hospitality group quoted on Primary Other

The agent's request. Please quote a Primary Other policy with a $5 million limit for a hospitality group, submitted through a wholesale broker with $150 million of annual revenue, with a $25,000 retention. The quote states an estimated premium of $329,175. It is ready to send.

Verdict: Blocked.

  • Segment not covered by this product. The insured segment is not one this Casualty product is described for (projects and practice products are for construction; Primary Other is for manufacturers and distributors; Excess Other is for real estate, hospitality, public entity and manufacturing).

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 quote for a wholesale broker on a casualty (general liability) policy: product, insured segment, revenue or project value, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the Bowhead Specialty Casualty product line as described in its public registration statement and returns a verdict, every reason, a line-item quote and the figures it computed.
  2. I will start with the first request by the agent.
  3. The agent says. Please quote a Primary Practice policy with a $10 million limit for a construction contractor, submitted through a retail agent with $80 million of annual revenue, with a $10,000 retention. The quote states an estimated premium of $772,800. The rest of the request is on screen.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with two findings and one approval, all in one answer. Casualty products are distributed only through wholesale brokers; a retail agent submission cannot be quoted. An insured with $80 million of revenue needs a retention of at least $25,000 (modeled).
  6. The validator returned two 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 quote are checked against each other.
  7. What this shows: for excess layers whose factor depends on the underlying limits, plus a minimum premium, a rules-based CPQ is error-prone for this.
  8. Now the agent resubmits.
  9. The corrected request. It changes the distribution channel to wholesale broker, the retention to $50,000 and the stated premium to $695,520.
  10. It calls the validator again.
  11. Not blocked, but held for a person.
  12. The validator also returns the line-item quote. It starts with Primary Practice base premium: annual revenue of $80 million at a typical rate per $1 million and Limit factor of 230 percent for a $10 million limit; the other line is on screen.
  13. The two blocking findings are cleared; one approval remains for a person: limit above the usual net retention.
  14. The agent resubmits once more.
  15. The revised request. It changes the limit to 5 million dollars and the stated premium to $529,200.
  16. It calls the validator once more.
  17. This time the validator clears it.
  18. The validator also returns the line-item quote. It starts with Primary Practice base premium: annual revenue of $80 million at a typical rate per $1 million and Limit factor of 175 percent for a $5 million limit; the other line is on screen.
  19. The earlier finding is cleared, and the validator's total, $529,200.00, matches the quote.
  20. Three problems caught in one call, fixed in two passes, and a clean $529,200 quote. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.