Predictika · Seat-and-Tier Deal Desk Check

Before an AI agent sends a software deal, the deal is checked against the vendor’s discount and approval rules

Orrin Software, a fictional software vendor, lets an AI deal-desk agent assemble multi-year seat deals across its Core, Insights and Copilot products. Before the agent sends one, the call goes to Predictika’s Seat-and-Tier Deal Desk Check. It applies the discount caps, counts the free months into the blended discount, checks the product and tier dependencies, the billing, partner and renewal terms, works out which approval level the deal needs, and checks that the discount and contract value the agent stated match the computed ones. Then it returns one verdict: allow, needs approval, or block.

Illustrative only. Orrin Software and its customers are fictional. Product names, prices, discount limits and approval levels are invented for this demonstration.

Scope: one multi-year seat deal, judged against discount, dependency, channel and approval rulesSee everything this validator covers

This validator decides one thing: whether one multi-year seat deal may be sent, given what is in it. It is not a pricing system, and it does not know the customer.

  • Actions — send one multi-year seat deal for Orrin Core, Orrin Insights and Orrin Copilot seats on the Team or Business tier, direct or through a partner.
  • Rules — Core discount cap and its competitive exception, add-on discounts, free months, the effective discount ceiling, product and tier dependencies, billing against discount, partner margin, deal registration, renewal cap, the approval level, and the stated discount and contract value.

Within these rules the validator handles any combination of the listed choices, not just the seven scenarios on this page. It never sees the customer, the contract or the order system; those facts are taken as stated.

What the validator sees: the deal the agent proposes, never the customer or the order systemRead more

The validator takes the deal as inputs: the seats on each product, the discounts, the term and free months, the tier, billing, channel, the clauses, the renewal cap, the sign-off on file, and the discount and contract value the agent stated. It computes the figures and applies the rules.

Two honest limits. The validator judges the deal it is given: if the agent gets a fact about the customer wrong, the answer follows it. And a deal that never calls the validator is not checked; in this demo the agent calls it itself, and in a deployment the deal workflow makes the call.

The AI agentProposes sending a deal, says why it is fine, then calls the validator before sending.
Predictika validatorReads the deal, computes the discount and contract value, applies the rules, then returns one verdict.

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 agent says

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

Proposing agent: AI agent (scripted)

Before sending, the agent must call Recorded run

Predictika validator


      

        
        
        
      

      
Waiting for the call.

      
    

  

  
  

    
Where quoting tools get it wrong
Click to expand
    
General statements about quoting tools, not about any named product. We rest this on public descriptions of how such tools work; anyone running a specific product will know it better than we do.

    
What the validator caughtTypical toolPredictika
Core discount cap with a competitive-displacement exception, and add-ons no deeper than CoreOnly if someone is vigilant
Discount caps are common in quoting tools. An exception that depends on the term and a flag, plus an ordering rule between product lines, has to be written as custom rules.		Every rule is checked on every call and the verdict lists each finding.
Blended discount that counts the free monthsCannot do it
A quote shows line discounts. Folding free months into the effective discount and testing it against a maximum needs a calculation that reads every line, the term and the free months.		Every rule is checked on every call and the verdict lists each finding.
Approval level set by discount depth, contract value and terms togetherOnly with workarounds
Approval ladders are usually kept per field. Taking the highest level across the discount, the value and the clauses, then comparing it with the sign-off on file, needs a rule that reads all three.		Every rule is checked on every call and the verdict lists each finding.
Checking the agent’s stated discount and contract valueOnly with workarounds
Many tools show a total. Recomputing it from the seats, discounts and free months and comparing it with what the agent stated needs a rule that does both.		Every rule is checked on every call and the verdict lists each finding.
Seat and tier prerequisites across productsOnly if someone is vigilant
Prerequisites live in price book notes. A rule has to apply the tier and seat-count conditions to each quote.		Every rule is checked on every call and the verdict lists each finding.
Core discount over the standard capOnly with workarounds
A quoting tool can hold a discount cap and an approval ladder. Folding the free months into the blended discount, and then checking the agent’s stated figures against that result, takes rules written to read them together.		Every rule is checked on every call and the verdict lists each finding.
Copilot seats without their prerequisitesOnly if someone is vigilant
Tier and seat dependencies are written in the price book notes. A rule has to read the tier and the three seat counts together.		Every rule is checked on every call and the verdict lists each finding.
Customer discount plus partner margin too deepOnly if someone is vigilant
Channel rules sit outside the quote. A rule has to add the customer discount and the partner margin and read the registration status.		Every rule is checked on every call and the verdict lists each finding.
Sign-off level above what is on fileOnly if someone is vigilant
The approval level depends on the discount, the contract value and the terms together. A rule has to take the highest of the three.		Every rule is checked on every call and the verdict lists each finding.

  

  

    
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 request; approval findings need a person to sign off first.

    
1The effective discount of 39.7% (including free months) is over the 38% company maximum that no approval level can authorize.
2A 30% Core discount is above the 22% standard cap; deeper requires a documented competitive displacement and a term of at least 3 years.
3Add-on discounts (30% Insights, 35% Copilot) may not be deeper than the 30% Core discount.
4Copilot seats need the Business tier and Insights seats at least equal to Copilot seats; this quote has the Team tier, 350 Insights seats and 200 Copilot seats.
5350 Insights seats exceed the 300 Core seats they sit on.
63 free months exceed the allowance for a 1-year term (2 months, or 5 months on terms of 3 years or more).
7Monthly billing is not allowed when the effective discount is 23.0%; deals discounted past 22% must bill annually upfront.
8The 15.0% customer discount plus a 40% partner margin stack past the 50% combined concession limit.
9This account has a conflicting channel deal registration; the partner quote cannot be issued until it is resolved.
10A 3% renewal price-protection cap is below the 4% floor the company allows.
11The stated blended discount of 30% does not match the computed effective discount of 36.3% (including free months).
12The stated contract value of $188145 does not match the computed net contract value of $172466.25.
13This deal needs CFO sign-off (discount depth, contract value and terms set the level) and only VP approval is on file.
14A most-favored-customer or uncapped-liability term needs Legal review before it is offered.
15A 5% renewal price-protection cap is below the standard 6%; confirm with revenue operations.

  

  

    
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. Discount depth: Deep discounts and figures that ignore the free months

    
The agent's request. I'm sending Northgate Logistics a 3-year deal on the Business tier: 500 Core seats at 30%, 300 Insights seats at 30% and 150 Copilot seats at 35%, with 3 free months. The blended discount is 30% and the contract value is 188145.

    
Verdict: Blocked.

    
    
      
  • Core discount over the standard cap. A 30% Core discount is above the 22% standard cap; deeper requires a documented competitive displacement and a term of at least 3 years.
    • 
      
  • Add-on discount deeper than the Core discount. Add-on discounts (30% Insights, 35% Copilot) may not be deeper than the 30% Core discount.
    • 
      
  • Stated discount does not match. The stated blended discount of 30% does not match the computed effective discount of 36.3% (including free months).
    • 
      
  • Stated contract value does not match. The stated contract value of $188145 does not match the computed net contract value of $172466.25.
    • 
      
  • Sign-off level above what is on file. This deal needs CFO sign-off (discount depth, contract value and terms set the level) and only VP approval is on file.
    • 
    

    
2. Discount depth: The same deal inside the discount rules

    
The agent's request. I'm sending Northgate Logistics the same 3-year Business deal, with 500 Core seats at 22%, 300 Insights seats at 20% and 150 Copilot seats at 22%, and 2 free months. The blended discount is 26% and the contract value is 200379.

    
Verdict: Allowed.

    
3. Product dependencies: Copilot on the Team tier, more Insights seats than Core seats

    
The agent's request. I'm sending Calder Biotech a 2-year deal on the Team tier: 300 Core seats, 350 Insights seats and 200 Copilot seats, all at 10%. The blended discount is 10% and the contract value is 125280.

    
Verdict: Blocked.

    
    
      
  • Copilot seats without their prerequisites. Copilot seats need the Business tier and Insights seats at least equal to Copilot seats; this quote has the Team tier, 350 Insights seats and 200 Copilot seats.
    • 
      
  • More Insights seats than Core seats. 350 Insights seats exceed the 300 Core seats they sit on.
    • 
    

    
4. Product dependencies: Business tier, Insights within Core

    
The agent's request. I'm sending the same deal on the Business tier, with 300 Core seats, 300 Insights seats and 200 Copilot seats, all at 10%. The blended discount is 10% and the contract value is 120960.

    
Verdict: Allowed.

    
5. Partner deal: A partner deal with a stacked margin

    
The agent's request. I'm sending Wexford Payments a 2-year partner deal on the Business tier: 400 Core seats and 200 Insights seats at 15%, with a 40% partner margin and a 5% renewal cap. The blended discount is 15% and the contract value is 106080.

    
Verdict: Blocked.

    
    
      
  • Customer discount plus partner margin too deep. The 15.0% customer discount plus a 40% partner margin stack past the 50% combined concession limit.
    • 
      
  • Conflicting deal registration. This account has a conflicting channel deal registration; the partner quote cannot be issued until it is resolved.
    • 
      
  • Renewal cap below the standard. A 5% renewal price-protection cap is below the standard 6%; confirm with revenue operations.
    • 
    

    
6. Partner deal: Margin lowered, registration cleared, one approval left

    
The agent's request. I'm sending the Wexford Payments partner deal again with a 30% partner margin and the deal registration conflict resolved, on the same 400 Core seats and 200 Insights seats at 15%, with a 5% renewal cap. The blended discount is 15% and the contract value is 106080.

    
Verdict: Needs approval.

    
    
      
  • Renewal cap below the standard. A 5% renewal price-protection cap is below the standard 6%; confirm with revenue operations.
    • 
    

    
7. Approval ladder: A deep three-year deal with a convenience clause

    
The agent's request. I'm sending Larkspur Foods a 3-year deal on the Business tier: 1200 Core seats at 18%, 600 Insights seats at 15% and 300 Copilot seats at 18%, with 3 free months, a termination-for-convenience clause and a 4% renewal cap. The blended discount is 24% and the contract value is 469161.

    
Verdict: Needs approval.

    
    
      
  • Sign-off level above what is on file. This deal needs VP sign-off (discount depth, contract value and terms set the level) and only Director approval is on file.
    • 
      
  • Renewal cap below the standard. A 4% renewal price-protection cap is below the standard 6%; confirm with revenue operations.
    • 
    

  

  

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

    
    
      
  1. Welcome to a demo of Predictika's Validator. Orrin Software, a fictional software vendor, lets an AI deal-desk agent assemble multi-year seat deals across its Core, Insights and Copilot products. Before the agent sends one, the call goes to Predictika’s Seat-and-Tier Deal Desk Check. It applies the discount caps, counts the free months into the blended discount, checks the product and tier dependencies, the billing, partner and renewal terms, works out which approval level the deal needs, and checks that the discount and contract value the agent stated match the computed ones. Then it returns one verdict: allow, needs approval, or block.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. I'm sending Northgate Logistics a 3-year deal on the Business tier: 500 Core seats at 30%, 300 Insights seats at 30% and 150 Copilot seats at 35%. The blended discount is 30% and the contract value is 188145. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with four findings and one approval, all in one answer. A 30% Core discount is above the 22% standard cap; deeper requires a documented competitive displacement and a term of at least 3 years. Add-on discounts (30% Insights, 35% Copilot) may not be deeper than the 30% Core discount. The stated blended discount of 30% does not match the computed effective discount of 36.3% (including free months). The other one is on screen.
    6. 
      
  6. The validator returned four 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 request are checked against each other.
    7. 
      
  7. What this shows: for sign-off level above what is on file, a tool does this reliably only when someone is vigilant.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the term to 3 years, 2 free months, the discounts to 22 / 20 / 22 percent, the stated blended discount to 26 percent and the stated contract value to 200379 USD.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. All five earlier findings are cleared.
    13. 
      
  13. Five problems caught in one call, fixed in one pass, and a clean request. That is the value of Predictika's validator: catching mistakes before the action runs.
    14.