Predictika · Printed circuit board and IC substrate quotes

An AI sales agent's PCB and IC substrate quote, checked before it goes out

An AI sales agent prepares a quote for printed circuit boards or IC substrates: the plant, the technology, the number of layers, the line/space, high-frequency material, the customer segment, the quantity, and a total. Before the quote goes out, the agent sends it to Predictika, which checks it against a declarative model of what each plant makes and returns a verdict, every reason and the totals it computed.

Scope: PCB and IC substrate quotes across seven plants and eight technologies: capability, layers, line/space, high-frequency material, segment focus, series size, order value and the stated total. See everything this validator covers

This validator models PCB and IC substrate quotes, not just the scenarios shown below.

  • Plant and technology — Leoben, Fehring, Chongqing, Shanghai, Nanjangud and Kulim against multilayer, HDI, any-layer HDI, substrate-like PCB, IC substrate, embedded component packaging, rigid-flex and insulated metal substrate; a technology a plant does not list is blocked.
  • Layers and line/space — each plant and technology has a maximum layer count and a finest line/space, and a quote beyond either is blocked.
  • High-frequency material — some plants list high-frequency boards up to a layer count, others do not list them at all.
  • Segment and series size — a segment outside the plant's listed focus, or a small series on a fine-pitch technology, needs a person to approve it.
  • Price and total — boards, tooling, volume tiers and the order total are computed, a total over EUR 1,000,000 needs approval, and the total the quote states is compared with the computed one.

Any PCB or IC substrate quote that touches these rules is in scope — not only the scenarios pictured here. Which plant makes which technology, and the plant focus, follow public AT&S information. Layer and line/space limits beyond those, all prices, tooling fees, volume tiers and the approval thresholds are modeled from comparable products and are illustrative, not published by AT&S.

The AI sales agentBuilds the quote from the conversation with the customer and says it is ready.
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 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
Plant and technology capabilityOnly with care
Which plant makes which technology is a two-way table that has to be kept in step with the catalog, and a pairing that is missing must be caught as an error rather than priced.		Every plant and technology pairing is checked in the same pass as all the other rules.
Layers and line/space limits that depend on the plantOnly with workarounds
The limits differ for every plant and technology, so a single rule per product does not work and the table of limits has to be maintained by hand.		Both limits are looked up from the plant and technology on every call, and each finding names the limit that was exceeded.
High-frequency material by plantOnly with care
Some plants list high-frequency boards up to a layer count and others do not list them, which is another plant-dependent limit.		The high-frequency limit is checked together with the layer count, and a plant that does not list it says so.
Segment focus and small seriesOnly with workarounds
These are approvals rather than errors, so they must be kept apart from blocking rules, and the small-series check applies only to fine-pitch technologies.		Each finding says whether it blocks or needs a person, and all of them come back together.
A stated total reconciled with the computed oneOnly with workarounds
The total is built from boards, volume tier and tooling, and a separate calculation is needed to compare it with the figure the quote states.		The total is computed inside the model and compared with the stated one on every call.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the quote is saved or a step is completed, so a person 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.

    
1The quote has 30 layers; this plant and technology are modeled up to 28 layers.
2The selected plant does not list this technology in its public capability; choose a plant that makes it.
3The quote asks for line/space of 3 um; this plant and technology are modeled down to 5 um.
4The quote needs high-frequency material on 12 layers; this plant lists high-frequency boards up to 10 layers (0 means not listed).
5The quote states a total of EUR 150,000; the quoted boards and tooling compute to EUR 413,000.
6The quote is for a segment that is not in the listed focus of this plant; a person should confirm the plant assignment.
7A series order of 300 boards on this technology is below the modeled minimum of 500; an engineering review is needed.
8The order total of EUR 3,929,000 is over EUR 1,000,000; sales management approval is needed.

  

  

    
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. Kulim IC substrate quote: IC substrate quote from Kulim, with several gaps

    
The agent's request. Please quote 400 boards in IC substrate from the Kulim plant, with 30 layers and a line/space of 3 um, in high-frequency material. The customer is in the mobile segment and the quote states a total of EUR 397,000. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Layer count over the plant limit. The quote has 30 layers; this plant and technology are modeled up to 26 layers.
    • 
      
  • Line/space finer than the plant can make. The quote asks for line/space of 3 um; this plant and technology are modeled down to 5 um.
    • 
      
  • High-frequency material beyond the plant. The quote needs high-frequency material on 30 layers; this plant lists high-frequency boards up to 0 layers (0 means not listed).
    • 
      
  • Segment outside the plant focus. The quote is for a segment that is not in the listed focus of this plant; a person should confirm the plant assignment.
    • 
      
  • Small series on a fine-pitch technology. A series order of 400 boards on this technology is below the modeled minimum of 500; an engineering review is needed.
    • 
    

    
2. Kulim IC substrate quote: The agent resubmits after fixing the errors

    
The agent's request. Please quote 400 boards in IC substrate from the Kulim plant, with 26 layers and a line/space of 5 um. The customer is in the mobile segment and the quote states a total of EUR 357,000. It is ready to send.

    
Verdict: Needs approval.

    
    
      
  • Segment outside the plant focus. The quote is for a segment that is not in the listed focus of this plant; a person should confirm the plant assignment.
    • 
      
  • Small series on a fine-pitch technology. A series order of 400 boards on this technology is below the modeled minimum of 500; an engineering review is needed.
    • 
      
  • Price summary. 400 boards at EUR 830 each = EUR 332,000, plus tooling EUR 25,000, total EUR 357,000 (modeled, illustrative prices).
    • 
    

    
3. Kulim IC substrate quote: The agent resubmits after fixing the errors

    
The agent's request. Please quote 1,000 boards in IC substrate from the Kulim plant, with 26 layers and a line/space of 5 um. The customer is in the server segment and the quote states a total of EUR 855,000. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 1,000 boards at EUR 830 each = EUR 830,000, plus tooling EUR 25,000, total EUR 855,000 (modeled, illustrative prices).
    • 
    

    
4. Insulated metal substrate quote: Insulated metal substrate quote from Leoben with a wrong total

    
The agent's request. Please quote 1,000 boards in insulated metal substrate from the Leoben plant, with 2 layers and a line/space of 100 um. The customer is in the industrial segment and the quote states a total of EUR 50,000. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Technology not offered at this plant. The selected plant does not list this technology in its public capability; choose a plant that makes it.
    • 
      
  • Stated total differs from the computed total. The quote states a total of EUR 50,000; the quoted boards and tooling compute to EUR 52,000.
    • 
    

    
5. Insulated metal substrate quote: The agent resubmits after fixing the errors

    
The agent's request. Please quote 1,000 boards in insulated metal substrate from the Fehring plant, with 2 layers and a line/space of 100 um. The customer is in the industrial segment and the quote states a total of EUR 52,000. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 1,000 boards at EUR 50 each = EUR 50,000, plus tooling EUR 2,000, total EUR 52,000 (modeled, illustrative prices).
    • 
    

    
6. Nanjangud HDI quote: HDI quote from Nanjangud for a mobile program

    
The agent's request. Please quote 1,000 boards in HDI from the Nanjangud plant, with 14 layers and a line/space of 40 um. The customer is in the mobile segment and the quote states a total of EUR 174,500. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Layer count over the plant limit. The quote has 14 layers; this plant and technology are modeled up to 12 layers.
    • 
      
  • Line/space finer than the plant can make. The quote asks for line/space of 40 um; this plant and technology are modeled down to 60 um.
    • 
      
  • Segment outside the plant focus. The quote is for a segment that is not in the listed focus of this plant; a person should confirm the plant assignment.
    • 
    

    
7. Nanjangud HDI quote: The agent resubmits after fixing the errors

    
The agent's request. Please quote 1,000 boards in HDI from the Nanjangud plant, with 12 layers and a line/space of 60 um. The customer is in the automotive segment and the quote states a total of EUR 156,500. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 1,000 boards at EUR 153 each = EUR 153,000, plus tooling EUR 3,500, total EUR 156,500 (modeled, illustrative prices).
    • 
    

  

  

    
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 printed circuit boards or IC substrates: the plant, the technology, the number of layers, and more. Before the quote goes out, the agent sends it to Predictika, which checks it against a declarative model of what each plant makes and returns a verdict, every reason and the totals it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote 400 boards in IC substrate from the Kulim plant, with 30 layers and a line/space of 3 um, in high-frequency material. The customer is in the mobile segment and the quote states a total of EUR 397,000. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with three findings and two approvals, all in one answer. The quote has 30 layers; this plant and technology are modeled up to 26 layers. The quote asks for line/space of 3 um; this plant and technology are modeled down to 5 um. The quote needs high-frequency material on 30 layers; this plant lists high-frequency boards up to 0 layers (0 means not listed).
    6. 
      
  6. The validator returned three 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 layers and line/space limits that depend on the plant, a rules-based CPQ can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the layers to 26, the line/space to 5 um, the high-frequency material to no and the total stated on the quote to 357,000.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Not blocked, but held for a person.
    12. 
      
  12. The three blocking findings are cleared; two approvals remain for a person: segment outside the plant focus and small series on a fine-pitch technology.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It changes the quantity to 1,000, the segment to server and the total stated on the quote to 855,000.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. Both earlier findings are cleared.
    18. 
      
  18. Five 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.