Predictika · HPE servers and racks

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

An AI sales agent prepares a quote for HPE ProLiant servers and racks: platform, processors, memory, GPUs, network adapters, drives, power supplies, rack power and space, cooling, support and deployment. Before it goes out, the agent sends it to Predictika's validator, which checks it against a declarative model of the platform and rack rules, compares the total the agent states with its own, and returns a verdict, every reason, and the figures it computed.

Scope: HPE ProLiant servers (DL360 and DL380 Gen12, DL385 Gen11, Compute XD685) and NVIDIA GB300 NVL72 by HPE: processors and sockets, power and cooling, GPUs, drives, rack power and space, facility fit, and the stated total. See everything this validator covers

This validator checks AI-drafted quotes for HPE ProLiant servers and racks, not just the scenarios shown below. It does not price or configure any other HPE product line.

  • Processor, memory and slots — processor socket and thermal rating against what each chassis supports, processor count against sockets, and DIMM count against the slots the installed processors give.
  • GPUs, drives and network adapters — GPU form factor, count and thermal rating, the fully populated baseboard on the 8-GPU platforms, drive form factor and bays, and adapter slots.
  • Power supplies against full draw — supply type and bay count, and the server's full draw against the usable supply capacity with the redundant spares held back.
  • Racks, facility and deployment — racks needed from rack power and rack units across every line, facility water for liquid cooling, site power per rack for rack-scale systems, and the deployment service that factory-integrated racks require.
  • Stated total — the total the agent states against the total the validator computes from list prices, the volume discount band, support and deployment.

Any HPE server and rack quote that touches these rules is in scope — not only the scenarios pictured here. Not covered: other HPE product lines, storage and networking beyond the top-of-rack switch, licenses, tax and freight. All prices, discount bands and support percentages are Predictika illustrative figures, not HPE list prices or HPE data.

The AI sales agentBuilds the server and rack 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.

8 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
Parts that must fit the platformOnly with care
Socket, thermal rating, DIMM slots, drive form and bays, GPU form and count are pairwise fits between two catalogs, so each usually becomes one rule per combination and the list grows with every new platform.		Socket, rating, slots and bays are properties of the catalog items, checked against the chosen platform on every call.
Power supplies against full draw, spares held backOnly with workarounds
The draw is a sum over processors, memory, GPUs, adapters and drives, and the usable supply is the installed supplies minus the spares, so the rule has to be rewritten per platform.		Draw and usable supply capacity are computed together, so a changed processor or GPU shows up in the power check in the same answer.
A fix to one part changes the nextOnly with workarounds
Swapping a GPU or processor changes the draw, the servers that fit in a rack, the rack count and the price; rules that fire at save or step time rarely re-check all of them together.		Every figure is recomputed on every call, so the resent quote is checked as a whole, not only where the agent changed it.
Racks across mixed server linesOnly with workarounds
Rack count depends on power and rack units summed over several server lines; price rules rarely see it.		The racks needed are derived from every line and compared with the racks the agent chose.
Facility facts: water, power per rack, deploymentOnly with care
Liquid cooling, rack-scale power and the deployment requirement depend on site facts that sit outside the product rules, so each needs its own conditional rule.		The site facts are fields of the same request and are evaluated with every other rule.
The total the agent statesOnly with workarounds
Comparing a stated figure with a computed one means a separate rule per layout, and price rules rarely run on stated figures.		The stated total is a field in the request and is checked against the computed total.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the quote is built, so a problem raised late can hide an earlier one.		Every rule is evaluated at the end of input and all findings come back together.

  

  

    
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.

    
1HPE ProLiant DL360 Gen12: Intel Xeon 6 6787P is rated 350 W; this chassis cools up to 300 W per processor.
2HPE ProLiant DL385 Gen11: Intel Xeon 6 6737P uses socket LGA4710 but this platform takes SP5.
3HPE ProLiant DL380 Gen12: NVIDIA H200 NVL is rated 600 W; this chassis cools up to 450 W per GPU.
4HPE ProLiant Compute XD685 air-cooled: NVIDIA HGX B300 288GB is rated 1,100 W; this chassis cools up to 1,000 W per GPU.
58 liquid-cooled servers are on the quote but the site has no facility water for liquid cooling.
6HPE ProLiant Compute XD685 direct liquid cooled: 4 GPUs chosen; the GPU baseboard comes fully populated with 8.
7HPE ProLiant Compute XD685 air-cooled: NVIDIA L40S is a PCIe GPU; this platform takes HGX or OAM 8-GPU baseboards.
8A rack-scale system draws 135 kW per rack; the site delivers 100 kW per rack.
9Rack-scale systems ship as factory-integrated racks and must be ordered with the deployment service.
10HPE ProLiant Compute XD685 direct liquid cooled: the server draws up to 11,260 W but its power supplies give 9,000 W with 2 kept as redundant spares.
111 rack chosen; these servers need at least 4 (up to 34.6 kW and 47 RU for servers per rack).
12HPE ProLiant DL360 Gen12: 40 DIMMs chosen but 2 processors give 32 slots.
13HPE ProLiant DL360 Gen12: HPE 3000W 54V Power Supply (S4R02A) does not fit this platform (it takes FlexSlot supplies).
14One HPE ProLiant Compute XD685 air-cooled draws 10.92 kW and is 6 RU; a rack here carries 8 kW and 47 RU for servers.
15The total the agent states ($611,804.00) differs from the total the validator computes ($583,699.20).
16HPE ProLiant DL380 Gen12: no processor chosen.
17HPE ProLiant DL380 Gen12: 3 processors chosen; the board has 2 sockets.
18HPE ProLiant Compute XD685 air-cooled: 1 processors chosen; this platform needs 2.
19HPE ProLiant Compute XD685 air-cooled: 24 DIMMs chosen but 1 processors give 12 slots.
20HPE ProLiant DL380 Gen12: no memory chosen.
21HPE ProLiant DL380 Gen12: each processor needs at least one DIMM (1 DIMMs for 2 processors).
22HPE ProLiant Compute XD685 air-cooled: this is a GPU platform; choose its GPU baseboard.
23HPE ProLiant DL360 Gen12: 4 network adapters chosen; the platform has 3 slots.
24HPE ProLiant DL380 Gen12: HPE 7.68TB NVMe E3.S is an E3.S drive; the bays take U.2 drives.
25HPE ProLiant DL360 Gen12: 11 drives chosen; the platform has 10 bays.
26HPE ProLiant DL380 Gen12: 3 power supplies chosen; the platform has 2 bays.
27Rack-scale systems are direct liquid cooled but the site has no facility water for liquid cooling.
28The total the agent states ($583,701.20) differs from the total the validator computes ($583,699.20).
29HPE ProLiant DL380 Gen12: 6 GPUs chosen; the platform has room for 4.
30HPE ProLiant DL380 Gen12: the server draws up to 3,052 W but its power supplies give 2,200 W with 1 kept as redundant spares.

  

  

    
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. Six holds on a mixed order: An order for 20 DL360 Gen12 servers and 4 liquid-cooled XD685 servers: a 350 W Xeon, 40 DIMMs, half a GPU baseboard on the XD685, one rack, and a total that leaves out support and deployment

    
The agent's request. Please quote 20 HPE ProLiant DL360 Gen12 servers, each with 2 processors (Intel Xeon 6 6787P), 40 DIMMs (HPE 32GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 2 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 4 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Racks: 1. Site: 60 kW per rack, facility water not available, deployment service ordered. Support: HPE Complete Care 3-year. The total is $1,894,308.00. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Processor runs hotter than the chassis cools. HPE ProLiant DL360 Gen12: Intel Xeon 6 6787P is rated 350 W; this chassis cools up to 300 W per processor.
    • 
      
  • More DIMMs than memory slots. HPE ProLiant DL360 Gen12: 40 DIMMs chosen but 2 processors give 32 slots.
    • 
      
  • Power supplies cannot carry the server. HPE ProLiant DL360 Gen12: the server draws up to 1,182 W but its power supplies give 1,000 W with 1 kept as redundant spares.
    • 
      
  • Partial GPU baseboard. HPE ProLiant Compute XD685 direct liquid cooled: 4 GPUs chosen; the GPU baseboard comes fully populated with 8.
    • 
      
  • Too few racks chosen. 1 rack chosen; these servers need at least 2 (up to 57.5 kW and 46 RU for servers per rack).
    • 
      
  • Liquid-cooled servers at a site without facility water. 4 liquid-cooled servers are on the quote but the site has no facility water for liquid cooling.
    • 
      
  • Stated total differs from the computed total. The total the agent states ($1,894,308.00) differs from the total the validator computes ($2,534,429.64).
    • 
    

    
2. Six holds on a mixed order: The agent resubmits after fixing the errors

    
The agent's request. Please quote 20 HPE ProLiant DL360 Gen12 servers, each with 2 processors (Intel Xeon 6 6737P), 32 DIMMs (HPE 32GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 2 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Racks: 1. Site: 60 kW per rack, facility water not available, deployment service ordered. Support: HPE Complete Care 3-year. The total is $1,894,308.00. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Too few racks chosen. 1 rack chosen; these servers need at least 2 (up to 57.5 kW and 46 RU for servers per rack).
    • 
      
  • Liquid-cooled servers at a site without facility water. 4 liquid-cooled servers are on the quote but the site has no facility water for liquid cooling.
    • 
      
  • Stated total differs from the computed total. The total the agent states ($1,894,308.00) differs from the total the validator computes ($2,986,736.04).
    • 
    

    
3. Six holds on a mixed order: The agent resubmits after fixing the errors

    
The agent's request. Please quote 20 HPE ProLiant DL360 Gen12 servers, each with 2 processors (Intel Xeon 6 6737P), 32 DIMMs (HPE 32GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 2 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Racks: 2. Site: 60 kW per rack, facility water available, deployment service ordered. Support: HPE Complete Care 3-year. The total is $3,181,413.72. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 24 servers in 2 racks plus 0 rack-scale systems; 64.2 kW. Hardware list $3,037,800.00, volume discount 22%, hardware net $2,369,484.00, support $781,929.72, deployment $30,000.00, total $3,181,413.72.
    • 
    

    
4. Liquid cooling and facility water: Eight liquid-cooled XD685 servers with B300 GPUs at a site with no facility water

    
The agent's request. Please quote 8 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Site: 60 kW per rack, facility water not available, deployment service ordered. Support: HPE Complete Care 3-year. The total is $4,732,326.72. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Liquid-cooled servers at a site without facility water. 8 liquid-cooled servers are on the quote but the site has no facility water for liquid cooling.
    • 
    

    
5. Liquid cooling and facility water: The agent resubmits after fixing the errors

    
The agent's request. Please quote 8 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Site: 60 kW per rack, facility water available, deployment service ordered. Support: HPE Complete Care 3-year. The total is $4,732,326.72. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 8 servers in 2 racks plus 0 rack-scale systems; 90.08 kW. Hardware list $4,532,800.00, volume discount 22%, hardware net $3,535,584.00, support $1,166,742.72, deployment $30,000.00, total $4,732,326.72.
    • 
    

    
6. Racks and power in a mixed order: Thirty DL380 servers and four XD685 servers squeezed into one rack, with five power supplies on each XD685

    
The agent's request. Please quote 30 HPE ProLiant DL380 Gen12 servers, each with 2 processors (Intel Xeon 6 6737P), 16 DIMMs (HPE 64GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 4 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 5 power supplies (the platform standard 3000 W supply). Racks: 1. Site: 40 kW per rack, facility water available, deployment service ordered. Support: HPE Tech Care Critical 5-year. The total is $3,469,105.20. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Power supplies cannot carry the server. HPE ProLiant Compute XD685 direct liquid cooled: the server draws up to 11,260 W but its power supplies give 9,000 W with 2 kept as redundant spares.
    • 
      
  • Too few racks chosen. 1 rack chosen; these servers need at least 4 (up to 34.6 kW and 47 RU for servers per rack).
    • 
    

    
7. Racks and power in a mixed order: The agent resubmits after fixing the errors

    
The agent's request. Please quote 30 HPE ProLiant DL380 Gen12 servers, each with 2 processors (Intel Xeon 6 6737P), 16 DIMMs (HPE 64GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 4 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 5 power supplies (the platform standard 3000 W supply). Racks: 4. Site: 40 kW per rack, facility water available, deployment service ordered. Support: HPE Tech Care Critical 5-year. The total is $3,839,084.40. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Power supplies cannot carry the server. HPE ProLiant Compute XD685 direct liquid cooled: the server draws up to 11,260 W but its power supplies give 9,000 W with 2 kept as redundant spares.
    • 
    

    
8. Racks and power in a mixed order: The agent resubmits after fixing the errors

    
The agent's request. Please quote 30 HPE ProLiant DL380 Gen12 servers, each with 2 processors (Intel Xeon 6 6737P), 16 DIMMs (HPE 64GB DDR5-6400 RDIMM), 1 network adapter (Broadcom 25Gb 2-port OCP3) and 4 drives (HPE 3.84TB NVMe SFF); plus 4 HPE ProLiant Compute XD685 direct liquid cooled servers, each with 2 processors (AMD EPYC 9655), 24 DIMMs (HPE 64GB DDR5-6400 RDIMM), 8 GPUs (NVIDIA HGX B300 288GB), 8 network adapters (NVIDIA ConnectX-8 800Gb), 4 drives (HPE 7.68TB NVMe E3.S) and 10 power supplies (the platform standard 3000 W supply). Racks: 4. Site: 40 kW per rack, facility water available, deployment service ordered. Support: HPE Tech Care Critical 5-year. The total is $3,863,108.40. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Price summary. 34 servers in 4 racks plus 0 rack-scale systems; 73.6 kW. Hardware list $3,482,700.00, volume discount 22%, hardware net $2,716,506.00, support $1,086,602.40, deployment $60,000.00, total $3,863,108.40.
    • 
    

  

  

    
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 HPE ProLiant servers and racks: platform, processors, memory, and more. Before it goes out, the agent sends it to Predictika's validator, which checks it against a declarative model of the platform and rack rules, compares the total the agent states with its own, and returns a verdict, every reason, and the figures it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote 20 HPE ProLiant DL360 Gen12 servers, each with 2 processors (Intel Xeon 6 6787P), 40 DIMMs (HPE 32GB DDR5-6400 RDIMM). Racks: 1. The rest of the request is on screen. The total is $1,894,308.00.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with seven findings, all in one answer. HPE ProLiant DL360 Gen12: Intel Xeon 6 6787P is rated 350 W; this chassis cools up to 300 W per processor. HPE ProLiant DL360 Gen12: 40 DIMMs chosen but 2 processors give 32 slots. HPE ProLiant DL360 Gen12: the server draws up to 1,182 W but its power supplies give 1,000 W with 1 kept as redundant spares. The other four are on screen.
    6. 
      
  6. The validator returned seven 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 and against the stated total.
    7. 
      
  7. What this shows: for facility facts: water, power per rack, deployment, a rules-based CPQ is error-prone for this.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the line 1 Processors to 2 x Intel Xeon 6 6737P, the line 1 Memory to 32 x HPE 32GB DDR5-6400 RDIMM and the line 2 GPUs to 8 x NVIDIA HGX B300 288GB.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Still blocked, with three findings, all in one answer. 1 rack chosen; these servers need at least 2 (up to 57.5 kW and 46 RU for servers per rack). 4 liquid-cooled servers are on the quote but the site has no facility water for liquid cooling. The total the agent states ($1,894,308.00) differs from the total the validator computes ($2,986,736.04).
    12. 
      
  12. Four findings cleared; still blocked on too few racks chosen, liquid-cooled servers at a site without facility water and stated total differs from the computed total.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It changes the racks to 2, the facility water to yes and the stated total to $3,181,413.72.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. The validator also returns the line-item quote. It starts with HPE ProLiant DL360 Gen12 and Intel Xeon 6 6737P; the other fifteen lines are on screen.
    18. 
      
  18. All three earlier findings are cleared, and the validator's total, $3,181,413.72, matches the quote.
    19. 
      
  19. Seven problems caught in one call, fixed in two passes, and a clean $3,181,413.72 quote. That is the value of Predictika's validator: catching mistakes before they reach the customer.
    20.