Predictika · Power Module Delivery Commitment Check

Before an AI agent commits a power module delivery, the quote is checked against the vendor’s capacity and supply rules

Kestrel Power Components, a fictional maker of DC-DC power modules, lets an AI quoting agent assemble module quotes and delivery commitments for customers who build them into arrays. Before the agent sends one, the call goes to Predictika’s Power Module Delivery Commitment Check. It checks the input and output voltage, derates the array for its cooling, checks the load, the transient peak and the case temperature, tests the order against the fab capacity and the customer allocation for the delivery period, applies the lead time, quote validity and NRE terms, prices the order, and checks that the capacity and total the agent stated match the computed ones. Then it returns one verdict: allow, needs approval, or block.

Illustrative only. Kestrel Power Components and its customers are fictional. Module models, ratings, prices, capacities and lead times are invented for this demonstration and are not technical or commercial guidance.

Scope: one module quote and delivery commitment, judged against capacity, supply and term rulesSee everything this validator covers

This validator decides one thing: whether one module quote and delivery commitment may be sent, given what is in it. It is not a design tool or a price book, and it does not know the customer.

  • Actions — send one quote for Kestrel modules, arranged in arrays and delivered in a chosen period.
  • Rules — voltage ranges, parallel limit, array capacity after cooling derate, transient peak, case temperature, fab capacity, customer allocation, lead time and expedite, quote validity, NRE, the sample limit, the price, and the stated figures.

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 design or the order system; those facts are taken as stated.

What the validator sees: the quote the agent proposes, never the design or the order systemRead more

The validator takes the quote as inputs: the module, the number per array, the cooling, the bus and output voltage, the load and peak, the case temperature, the order size, the delivery period, the allocation left, the requested lead time, the quote age, the part type and NRE status, the stage, and the capacity and total the agent stated. It computes the figures and applies the rules.

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

The AI agentProposes sending a quote, says why it is fine, then calls the validator before sending.
Predictika validatorReads the quote, derates the array, prices the order, 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
Array capacity from the module rating and the cooling methodCannot do it
Capacity depends on the rating, a cooling derate and the number of modules in parallel. Chaining those into one comparison with the load usually takes several rules that must run in order.		Every rule is checked on every call and the verdict lists each finding.
Fab capacity and customer allocation by periodOnly with workarounds
The units an order needs depend on modules per array and fab units per module, and the supply depends on the delivery period. A rules-based quoting tool usually needs a table kept by hand.		Every rule is checked on every call and the verdict lists each finding.
Volume price breaks and the expedite chargeOnly with workarounds
The unit price steps with the total module count and the expedite charge depends on the lead time. Both feed the total that the quote states.		Every rule is checked on every call and the verdict lists each finding.
Lead time, quote validity and NRE terms for custom partsOnly if someone is vigilant
Each is a simple validation rule. Keeping them complete as part types and stages are added is the work.		Every rule is checked on every call and the verdict lists each finding.
Checking the figures the agent statedCannot do it
Comparing a stated capacity and a stated total with computed ones needs rules that do both, and price rules rarely run on stated figures.		Every rule is checked on every call and the verdict lists each finding.
Load above the cooled array capacityOnly with workarounds
A quoting tool can look up a module rating. Derating it for the cooling method, comparing it with the load, then checking both stated figures takes rules written to read them together.		Every rule is checked on every call and the verdict lists each finding.
Order needs more fab units than are unallocatedOnly with workarounds
Fab capacity per period and customer allocation live in different places. A rule has to read the order, the period and the allocation together.		Every rule is checked on every call and the verdict lists each finding.
Lead time below the minimumOnly if someone is vigilant
Lead time, quote validity and NRE status sit in three different places. A rule has to read them together before a quote goes out.		Every rule is checked on every call and the verdict lists each finding.
Load close to the array capacityOnly if someone is vigilant
The thermal margin, the headroom margin and the cost review are separate approvals. A rule has to raise each one, so a person sees all 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.

    
1A 35 V input bus is outside the 36 to 60 V input range of the KD36 48V to 12V converter.
2A 9.4 V output is outside the 9.5 to 13.0 V output range of the KD36 48V to 12V converter.
3An array of 5 units is above the 4 unit parallel limit for the KD36 48V to 12V converter.
4The 2850 W load is above the 2808 W this array can deliver with forced air cooling.
5A transient peak of 5401 W is above the 5400.0 W peak the array can supply for 2 ms.
6A case temperature of 121 C is above the 120 C case limit.
7This order needs 32000 fab units and only 18000 are unallocated in Q1 2027; the delivery cannot be promised.
88000 modules is above the 7000 modules of allocation the customer has left; allocation must be released first.
914 weeks is below the 16 week minimum even with expedite for the KB62 bus converter.
10The quote is 75 days old and its validity is 60 days; re-price it.
11A customer-specific part needs its design-in NRE paid or scheduled before production quantities are quoted; the NRE is not paid.
12122 modules is above the 120 module limit for sample-stage orders.
13The quote states an array capacity of 3600 W but the validator computes 2808 W.
14The quote states a total of $72000 but the validator computes $79200.
15The 3000 W load uses more than 80 percent of the 3600 W array capacity; confirm the thermal and headroom margin.
16A case temperature of 110 C is above the 100 C design margin; confirm the thermal design.
1720 weeks is below the 24 week standard lead time for the KB62 bus converter; expedite needs approval.
18Component cost has moved 9 percent since the quote; honoring it needs a margin review.
19This order leaves under 2800 fab units of margin in Q4 2026; confirm with operations planning.
20This order uses more than 88 percent of the customer remaining allocation of 227 modules; confirm with account management.
21The NRE for this customer-specific part is pending; release production quantities only on written NRE terms.

  

  

    
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. Array capacity: A forced-air array that cannot carry the load

    
The agent's request. I'm sending Fenwick Compute a quote for 300 arrays of 2 KD36 converters on forced air cooling, for a 2850 W load with a 5200 W transient peak. The quote states an array capacity of 3600 W and a total of 72000.

    
Verdict: Blocked.

    
    
      
  • Load above the cooled array capacity. The 2850 W load is above the 2808 W this array can deliver with forced air cooling.
    • 
      
  • Stated capacity differs from the computed one. The quote states an array capacity of 3600 W but the validator computes 2808 W.
    • 
      
  • Stated total differs from the computed one. The quote states a total of $72000 but the validator computes $79200.
    • 
    

    
2. Array capacity: The same order on cold plate cooling

    
The agent's request. I'm sending Fenwick Compute the same order on cold plate cooling. The quote states an array capacity of 3600 W and a total of 79200.

    
Verdict: Allowed.

    
3. Supply commitments: A program larger than the allocation

    
The agent's request. I'm sending Oakmere Systems a quote for 4000 arrays of 2 KB62 converters, delivered in the first quarter of next year, for a 3400 W load. The customer has 7000 modules of allocation left. The quote states 4400 W and a total of 1984000.

    
Verdict: Blocked.

    
    
      
  • Order needs more fab units than are unallocated. This order needs 32000 fab units and only 18000 are unallocated in Q1 2027; the delivery cannot be promised.
    • 
      
  • Order above the customer allocation. 8000 modules is above the 7000 modules of allocation the customer has left; allocation must be released first.
    • 
    

    
4. Supply commitments: The program split to fit

    
The agent's request. I'm sending Oakmere Systems the same order split to 1500 arrays. The quote states 4400 W and a total of 810000.

    
Verdict: Allowed.

    
5. Custom parts: A custom part, an old quote and a short lead time

    
The agent's request. I'm sending Marren Instruments a quote for 1200 arrays of 2 KB62 converters on a custom part, for a 3400 W load, with delivery in 14 weeks. The quote is 75 days old and the design-in NRE is not paid. The quote states 4400 W and a total of 648000.

    
Verdict: Blocked.

    
    
      
  • Lead time below the minimum. 14 weeks is below the 16 week minimum even with expedite for the KB62 bus converter.
    • 
      
  • Quote older than its validity. The quote is 75 days old and its validity is 60 days; re-price it.
    • 
      
  • Custom part without its NRE. A customer-specific part needs its design-in NRE paid or scheduled before production quantities are quoted; the NRE is not paid.
    • 
    

    
6. Custom parts: Re-priced with a longer lead time

    
The agent's request. I'm sending Marren Instruments the same order re-priced: delivery in 20 weeks, a quote 20 days old, and the NRE scheduled on written terms. The quote states 4400 W and a total of 705600.

    
Verdict: Needs approval.

    
    
      
  • Lead time inside the expedite window. 20 weeks is below the 24 week standard lead time for the KB62 bus converter; expedite needs approval.
    • 
      
  • NRE still pending. The NRE for this customer-specific part is pending; release production quantities only on written NRE terms.
    • 
    

    
7. Approval authority: A hot enclosure and a cost increase

    
The agent's request. I'm sending Tamsin Controls a quote for 800 arrays of 4 KD18 converters, for a 3000 W load in an enclosure that reaches 110 C case temperature. Component cost has moved 9 percent since the quote. The quote states 3600 W and a total of 259200.

    
Verdict: Needs approval.

    
    
      
  • Load close to the array capacity. The 3000 W load uses more than 80 percent of the 3600 W array capacity; confirm the thermal and headroom margin.
    • 
      
  • Case temperature above the design margin. A case temperature of 110 C is above the 100 C design margin; confirm the thermal design.
    • 
      
  • Component cost moved since the quote. Component cost has moved 9 percent since the quote; honoring it needs a margin review.
    • 
    

  

  

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

    
    
      
  1. Welcome to a demo of Predictika's Validator. Kestrel Power Components, a fictional maker of DC-DC power modules, lets an AI quoting agent assemble module quotes and delivery commitments for customers who build them into arrays. Before the agent sends one, the call goes to Predictika’s Power Module Delivery Commitment Check. It checks the input and output voltage, derates the array for its cooling, checks the load, the transient peak and the case temperature, tests the order against the fab capacity and the customer allocation for the delivery period, applies the lead time, quote validity and NRE terms, prices the order, and checks that the capacity and total 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 Fenwick Compute a quote for 300 arrays of 2 KD36 converters on forced air cooling, for a 2850 W load with a 5200 W transient peak. The quote states an array capacity of 3600 W and a total of 72000.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with three findings, all in one answer. The 2850 W load is above the 2808 W this array can deliver with forced air cooling. The quote states an array capacity of 3600 W but the validator computes 2808 W. The quote states a total of $72000 but the validator computes $79200.
    6. 
      
  6. The validator returned three blocking findings 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 load above the cooled array capacity, a tool can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the cooling to cold plate and the stated quote total to 79200 USD.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. All three earlier findings are cleared.
    13. 
      
  13. Three 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.