Predictika · Server configuration quotes

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

An AI sales agent prepares a quote for MiTAC and TYAN servers: the server model, the number of chassis and nodes, the CPUs and memory modules, the GPUs, the U.2 drives, the power supplies, and the total memory and GPU counts it states. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the configuration limits and returns a verdict, every reason and the totals it computed.

Scope: MiTAC and TYAN server quotes: G8825Z5, G4520G6, TN85-B8261, GC68C-B8056, R2520G6, R2513G6 and M2810Z5, with nodes, CPUs, memory, GPUs, U.2 drives, power supplies and the stated memory and GPU totals. See everything this validator covers

This validator models server quotes of this kind, not just the scenarios shown below.

  • Model limits — seven server models and, for each, the nodes per chassis, CPU sockets, memory slots, GPUs and U.2 drive bays it holds, and the largest memory module modeled.
  • Memory and power — at least one memory module per CPU, the memory a customer needs per node against the quoted modules, redundant power supplies for GPU chassis and four supplies above four GPUs.
  • Totals and approvals — the total memory and GPU counts the quote states against the counts computed from the configuration, 256 GB modules and single power supplies that a person confirms.

Any server quote that touches these rules is in scope — not only the scenarios pictured here. Model names, form factors and the GPU, memory and drive counts follow public MiTAC and TYAN announcements where stated; the remaining limits and the power rules are modeled from comparable servers and not published by MiTAC. The validator models no prices, so it produces no price build-up. Not covered: other MiTAC and TYAN products, processors and GPU models, networking, software and warranties.

The AI sales agentBuilds the server quote from the conversation 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
Limits that differ for every server modelOnly with care
Each model has its own nodes, sockets, memory slots, GPU count and drive bays, so every limit rule has to read the model first.		The model and every count in the quote are checked together on every call.
Memory needed per node against the modules quotedOnly with workarounds
The memory a node gets is the number of modules times their size, and it has to be compared with the memory the customer needs.		The module count, module size and need are computed together on every call.
Power supplies that depend on the GPU countOnly with care
GPU chassis need two supplies and more than four GPUs need four, so a power rule has to read the GPU count.		The GPU count and the power supplies are checked together on every call.
Totals across chassis and nodesOnly with workarounds
Total memory and GPUs multiply chassis, nodes, modules and GPUs, so they must be computed before they can be compared with the totals stated.		The stated totals are fields of the quote and are compared with the computed totals 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 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.

    
1The quote has 9 GPUs per chassis; this model supports at most 8.
2The quote has 5 GPUs per chassis; this model supports at most 4.
3A chassis with more than 4 GPUs needs 4 power supplies; the quote has 2.
4The quote has 1 GPUs per chassis; this model supports at most 0.
5The quote has 13 U.2 drives per chassis; this model has 12 bays.
6The quote has 2 CPUs per node; this model has 1 socket(s) per node.
7The quote has 9 DIMMs per node; this model has 8 DIMM slots per node.
8The quote has 3 U.2 drives per chassis; this model has 2 bays.
9The quote has 5 nodes per chassis; this model holds at most 4.
10The quote has 2 nodes per chassis; this model holds at most 1.
11The customer needs 1,100 GB per node; the quoted DIMMs give 1,024 GB.
12The quote states 1,024 GB of memory in total; the quoted DIMMs give 2,048 GB.
13The quote states 8 GPUs in total; the quoted chassis give 16.
14The quote uses 256 GB DIMMs; a person should confirm availability and lead time.
15The largest DIMM modeled is 256 GB; the quote has 512 GB DIMMs.
16The quote has a single power supply and no power redundancy; a person should confirm the customer accepts this.
17A chassis with GPUs needs at least 2 power supplies; the quote has 1.
18Every CPU needs at least one DIMM; the quote has 1 DIMMs for 2 CPUs.

  

  

    
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. Small server limits: An R2513G6 quote with two CPUs, 12 memory modules, four GPUs, six U.2 drives and one power supply

    
The agent's request. Please quote 2 R2513G6 chassis, each node with 2 CPUs and 12 memory modules of 128 GB, 4 GPUs per chassis, 6 U.2 drives per chassis and 1 power supplies per chassis. The customer needs 2,048 GB of memory per node. The quote states 3,072 GB of memory and 8 GPUs in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Too many CPUs. The quote has 2 CPUs per node; this model has 1 socket(s) per node.
    • 
      
  • Too many DIMMs. The quote has 12 DIMMs per node; this model has 8 DIMM slots per node.
    • 
      
  • Memory needed exceeds configuration. The customer needs 2,048 GB per node; the quoted DIMMs give 1,536 GB.
    • 
      
  • Too many GPUs. The quote has 4 GPUs per chassis; this model supports at most 0.
    • 
      
  • Too many U.2 drives. The quote has 6 U.2 drives per chassis; this model has 2 bays.
    • 
      
  • GPU server needs redundant power. A chassis with GPUs needs at least 2 power supplies; the quote has 1.
    • 
    

    
2. Small server limits: The agent resubmits after fixing the errors

    
The agent's request. Please quote 2 R2513G6 chassis, each node with 1 CPU and 8 memory modules of 128 GB, 0 GPUs per chassis, 2 U.2 drives per chassis and 2 power supplies per chassis. The customer needs 512 GB of memory per node. The quote states 2,048 GB of memory and 0 GPUs in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 2 chassis, 2,048 GB of memory in total, 0 GPUs in total.
    • 
    

    
3. GPU server memory and power: A G8825Z5 quote with 288 GB memory modules, nine GPUs and two power supplies

    
The agent's request. Please quote 2 G8825Z5 chassis, each node with 2 CPUs and 24 memory modules of 288 GB, 9 GPUs per chassis, 8 U.2 drives per chassis and 2 power supplies per chassis. The customer needs 1,024 GB of memory per node. The quote states 13,824 GB of memory and 16 GPUs in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • DIMM size above 256 GB. The largest DIMM modeled is 256 GB; the quote has 288 GB DIMMs.
    • 
      
  • Too many GPUs. The quote has 9 GPUs per chassis; this model supports at most 8.
    • 
      
  • More than 4 GPUs needs 4 power supplies. A chassis with more than 4 GPUs needs 4 power supplies; the quote has 2.
    • 
      
  • Stated GPUs do not match. The quote states 16 GPUs in total; the quoted chassis give 18.
    • 
    

    
4. GPU server memory and power: The agent resubmits after fixing the errors

    
The agent's request. Please quote 2 G8825Z5 chassis, each node with 2 CPUs and 24 memory modules of 128 GB, 8 GPUs per chassis, 8 U.2 drives per chassis and 4 power supplies per chassis. The customer needs 1,024 GB of memory per node. The quote states 6,144 GB of memory and 16 GPUs in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 2 chassis, 6,144 GB of memory in total, 16 GPUs in total.
    • 
    

    
5. Multi-node server: An M2810Z5 quote with six nodes, 256 GB modules, 30 U.2 drives and one power supply

    
The agent's request. Please quote 2 M2810Z5 chassis with 6 nodes each, each node with 1 CPU and 8 memory modules of 256 GB, 0 GPUs per chassis, 30 U.2 drives per chassis and 1 power supplies per chassis. The customer needs 512 GB of memory per node. The quote states 24,576 GB of memory and 0 GPUs in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Too many nodes. The quote has 6 nodes per chassis; this model holds at most 4.
    • 
      
  • Too many U.2 drives. The quote has 30 U.2 drives per chassis; this model has 26 bays.
    • 
      
  • 256 GB DIMMs to confirm. The quote uses 256 GB DIMMs; a person should confirm availability and lead time.
    • 
      
  • Single power supply. The quote has a single power supply and no power redundancy; a person should confirm the customer accepts this.
    • 
    

    
6. Multi-node server: The agent resubmits after fixing the errors

    
The agent's request. Please quote 2 M2810Z5 chassis with 4 nodes each, each node with 1 CPU and 8 memory modules of 128 GB, 0 GPUs per chassis, 24 U.2 drives per chassis and 2 power supplies per chassis. The customer needs 512 GB of memory per node. The quote states 8,192 GB of memory and 0 GPUs in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 2 chassis, 8,192 GB of memory in total, 0 GPUs in total.
    • 
    

  

  

    
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 MiTAC and TYAN servers: the server model, the number of chassis and nodes, the CPUs and memory modules, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the configuration limits 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 2 R2513G6 chassis, each node with 2 CPUs and 12 memory modules of 128 GB, 4 GPUs per chassis. The customer needs 2,048 GB of memory per node. The rest of the request is on screen. The quote states 3,072 GB of memory and 8 GPUs in total.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with six findings, all in one answer. The quote has 2 CPUs per node; this model has 1 socket(s) per node. The quote has 12 DIMMs per node; this model has 8 DIMM slots per node. The customer needs 2,048 GB per node; the quoted DIMMs give 1,536 GB. The other three are on screen.
    6. 
      
  6. The validator returned six 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 quote are checked against each other.
    7. 
      
  7. What this shows: for memory needed per node against the modules quoted, a rules-based CPQ can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the CPUs per node to 1, the memory modules per node to 8, the GPUs per chassis to 0, the u.2 drives per chassis to 2, the stated total memory to 2,048 and the Stated total GPUs to 0. Two more changes are on screen.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. All six earlier findings are cleared.
    13. 
      
  13. Six problems caught in one call, fixed in one pass, 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.
    14.