Predictika · Data streaming cluster quotes

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

An AI sales agent prepares a quote for a managed Kafka cluster in Confluent Cloud: the cluster type, the cloud, the networking, the uptime SLA, the number of capacity units and the workload the cluster has to carry. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the published cluster types and limits and returns a verdict, every reason and the capacity figures it computed.

Scope: Confluent Cloud Kafka clusters (Basic, Standard, Enterprise, Freight, Dedicated): size limits, cloud, networking, uptime SLA and workload capacity. See everything this validator covers

This validator models Kafka cluster quotes for Confluent Cloud as published on the Confluent Cloud pricing page and in the cluster-types documentation, not just the scenarios shown below.

  • Cluster types and size — Basic (up to 50 eCKU), Standard (up to 10), Enterprise (up to 32), Freight (2 to 152, select AWS regions only) and Dedicated (up to 252 CKU on AWS, 152 on Google Cloud, 100 on Azure). A quote outside the range for its type is blocked.
  • Networking and uptime SLA — Basic and Standard clusters use public networking, Enterprise and Freight clusters private networking; the SLA offered depends on the type (Basic 99.5%, Standard and Enterprise 99.9% or 99.99%, Dedicated 99.95% single-zone or 99.99% multi-zone, Freight 99.99%), and the 99.99% SLA on Standard or Enterprise needs at least 2 eCKU.
  • Workload capacity — ingress and egress throughput, partitions and client connections are compared with what the chosen number of units supports, up to the cluster maximum for its type, and a capacity figure stated in the quote is compared with the one computed. Dedicated sizing, the Freight region and any commitment discount go to a person.

Any Confluent Cloud Kafka cluster quote that touches these rules is in scope — not only the scenarios pictured here. Not covered: prices (several rates are published only as ranges and Dedicated is quoted by sales, so this validator produces no price build-up or line-item quote), data transfer and storage charges, Connect, Flink, Stream Governance and support plans, Confluent Platform (self-managed), and per-region availability. The limits are those shown in the documentation read in October 2026 and may change.

The AI sales agentBuilds the Kafka cluster 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.

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
Limits that differ by cluster type and, for Dedicated, by cloudOnly with care
A product rule per type and cloud. It works, but the limits sit in a table that has to be kept in step with the documentation.		The type and cloud set the limits and every one is checked together.
Capacity that scales with the number of units, up to a cluster maximumOnly with care
A formula per type for each of throughput, partitions and connections, each capped at the maximum. It works, but there are many of them to maintain.		Each capacity is computed from the units and compared with the workload on every call.
An SLA that depends on the type, the zones and the number of unitsOnly with workarounds
The allowed SLA depends on three choices at once, so it takes cross-field rules that interact with each other.		The type, the zones, the SLA and the units are in one request and are checked together.
A stated capacity that disagrees with the published per-unit figuresOnly with workarounds
The figure the agent states is free text or a separate field, so it takes an extra rule that compares it with a computed value.		The stated figure is part of the request and is compared with the computed capacity 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.

    
1Dedicated clusters are pre-provisioned capacity whose specifications and price come from Confluent sales; a person should confirm the CKU count.
2Freight clusters are available in select AWS regions only; a person should confirm the region.
3The quote has 1 capacity units; this cluster type and cloud allows 2 to 152.
4Freight clusters are available in select AWS regions only; the quote is for another cloud.
5The quote has 11 capacity units; this cluster type and cloud allows 1 to 10.
6The quote has 51 capacity units; this cluster type and cloud allows 1 to 50.
7The quote has 33 capacity units; this cluster type and cloud allows 1 to 32.
8The quote has 253 capacity units; this cluster type and cloud allows 1 to 252.
9The quote has 153 capacity units; this cluster type and cloud allows 1 to 152.
10The quote has 101 capacity units; this cluster type and cloud allows 1 to 100.
11The quote has 0 capacity units; this cluster type and cloud allows 1 to 10.
12The workload peaks at 80 MBps of ingress; the chosen cluster size supports 0 MBps.
13The workload peaks at 200 MBps of egress; the chosen cluster size supports 0 MBps.
14The workload needs 800 partitions; the chosen cluster size supports 0.
15The workload has 3,000 client connections; the chosen cluster size supports 0.
16The networking choice is not supported by this cluster type; Basic and Standard clusters use public networking, Enterprise and Freight clusters use private networking.
17The chosen uptime SLA is not offered for this cluster type (Basic 99.5%; Standard and Enterprise 99.9% or 99.99%; Dedicated 99.95% single-zone or 99.99% multi-zone; Freight 99.99%).
18The quote asks for the 99.99% SLA with 1 eCKU; Standard and Enterprise clusters need at least 2 eCKU for it.
19The workload peaks at 101 MBps of ingress; the chosen cluster size supports 100 MBps.
20The workload peaks at 301 MBps of egress; the chosen cluster size supports 300 MBps.
21The workload needs 1,001 partitions; the chosen cluster size supports 1,000.
22The workload has 4,001 client connections; the chosen cluster size supports 4,000.
23The workload needs 60,001 partitions; the chosen cluster size supports 50,000.
24The quote states an ingress capacity of 250 MBps; the published per-unit capacity gives 100 MBps.

  

  

    
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. Standard to Enterprise: A Standard cluster with 12 units, private networking and a 99.99% SLA

    
The agent's request. Please quote a Standard Confluent Cloud Kafka cluster on AWS with 12 eCKU, private networking and a 99.99% uptime SLA. The workload peaks at 400 MBps of ingress and 1,100 MBps of egress, with 2,400 partitions and 11,000 client connections. The quote states an ingress capacity of 300 MBps. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Cluster size outside the allowed range. The quote has 12 capacity units; this cluster type and cloud allows 1 to 10.
    • 
      
  • Networking not supported by the cluster type. The networking choice is not supported by this cluster type; Basic and Standard clusters use public networking, Enterprise and Freight clusters use private networking.
    • 
      
  • Ingress throughput over capacity. The workload peaks at 400 MBps of ingress; the chosen cluster size supports 250 MBps.
    • 
      
  • Egress throughput over capacity. The workload peaks at 1,100 MBps of egress; the chosen cluster size supports 750 MBps.
    • 
      
  • Client connections over the cluster limit. The workload has 11,000 client connections; the chosen cluster size supports 10,000.
    • 
      
  • Stated capacity does not match. The quote states an ingress capacity of 300 MBps; the published per-unit capacity gives 250 MBps.
    • 
    

    
2. Standard to Enterprise: The agent resubmits after fixing the errors

    
The agent's request. Please quote a Enterprise Confluent Cloud Kafka cluster on AWS with 12 eCKU, private networking and a 99.99% uptime SLA. The workload peaks at 400 MBps of ingress and 1,100 MBps of egress, with 2,400 partitions and 11,000 client connections. The quote states an ingress capacity of 720 MBps. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Cluster summary. 12 capacity unit(s): ingress capacity 720 MBps, egress capacity 2,160 MBps, 36,000 partitions, 216,000 client connections.
    • 
    

    
3. Freight cluster: A Freight cluster on Azure with one unit, public networking and a 99.9% SLA

    
The agent's request. Please quote a Freight Confluent Cloud Kafka cluster on Azure with 1 eCKU, public networking and a 99.9% uptime SLA. The workload peaks at 50 MBps of ingress and 150 MBps of egress, with 2,000 partitions and 10,000 client connections. The quote states an ingress capacity of 60 MBps. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Cluster size outside the allowed range. The quote has 1 capacity units; this cluster type and cloud allows 2 to 152.
    • 
      
  • Freight only on AWS. Freight clusters are available in select AWS regions only; the quote is for another cloud.
    • 
      
  • Networking not supported by the cluster type. The networking choice is not supported by this cluster type; Basic and Standard clusters use public networking, Enterprise and Freight clusters use private networking.
    • 
      
  • Uptime SLA not offered. The chosen uptime SLA is not offered for this cluster type (Basic 99.5%; Standard and Enterprise 99.9% or 99.99%; Dedicated 99.95% single-zone or 99.99% multi-zone; Freight 99.99%).
    • 
    

    
4. Freight cluster: The agent resubmits after fixing the errors

    
The agent's request. Please quote a Freight Confluent Cloud Kafka cluster on AWS with 20 eCKU, private networking and a 99.99% uptime SLA. The workload peaks at 50 MBps of ingress and 150 MBps of egress, with 2,000 partitions and 10,000 client connections. The quote states an ingress capacity of 1,200 MBps. It is ready to send.

    
Verdict: Needs approval.

    
    
      
  • Freight region to confirm. Freight clusters are available in select AWS regions only; a person should confirm the region.
    • 
      
  • Cluster summary. 20 capacity unit(s): ingress capacity 1,200 MBps, egress capacity 3,600 MBps, 50,000 partitions, 360,000 client connections.
    • 
    

    
5. Basic to Standard: A Basic cluster with 60 units, private networking and a stated capacity of 300 MBps

    
The agent's request. Please quote a Basic Confluent Cloud Kafka cluster on Google Cloud with 60 eCKU, private networking and a 99.9% uptime SLA. The workload peaks at 200 MBps of ingress and 600 MBps of egress, with 1,200 partitions and 900 client connections. The quote states an ingress capacity of 300 MBps. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Cluster size outside the allowed range. The quote has 60 capacity units; this cluster type and cloud allows 1 to 50.
    • 
      
  • Networking not supported by the cluster type. The networking choice is not supported by this cluster type; Basic and Standard clusters use public networking, Enterprise and Freight clusters use private networking.
    • 
      
  • Uptime SLA not offered. The chosen uptime SLA is not offered for this cluster type (Basic 99.5%; Standard and Enterprise 99.9% or 99.99%; Dedicated 99.95% single-zone or 99.99% multi-zone; Freight 99.99%).
    • 
      
  • Stated capacity does not match. The quote states an ingress capacity of 300 MBps; the published per-unit capacity gives 250 MBps.
    • 
    

    
6. Basic to Standard: The agent resubmits after fixing the errors

    
The agent's request. Please quote a Standard Confluent Cloud Kafka cluster on Google Cloud with 8 eCKU, public networking and a 99.9% uptime SLA. The workload peaks at 200 MBps of ingress and 600 MBps of egress, with 1,200 partitions and 900 client connections. The quote states an ingress capacity of 200 MBps. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Cluster summary. 8 capacity unit(s): ingress capacity 200 MBps, egress capacity 600 MBps, 2,000 partitions, 8,000 client connections.
    • 
    

    
7. Dedicated and commitment: A Dedicated cluster on Azure with 120 units, a single-zone 99.99% SLA and an annual commitment

    
The agent's request. Please quote a Dedicated Confluent Cloud Kafka cluster on Azure with 120 CKU, private networking and a 99.99% uptime SLA, in a single zone. The workload peaks at 5,000 MBps of ingress and 14,000 MBps of egress, with 80,000 partitions and 1,500,000 client connections. The customer wants an annual commitment. The quote states an ingress capacity of 7,200 MBps. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Cluster size outside the allowed range. The quote has 120 capacity units; this cluster type and cloud allows 1 to 100.
    • 
      
  • Uptime SLA not offered. The chosen uptime SLA is not offered for this cluster type (Basic 99.5%; Standard and Enterprise 99.9% or 99.99%; Dedicated 99.95% single-zone or 99.99% multi-zone; Freight 99.99%).
    • 
      
  • Dedicated capacity to confirm. Dedicated clusters are pre-provisioned capacity whose specifications and price come from Confluent sales; a person should confirm the CKU count.
    • 
      
  • Commitment discount to confirm. A commitment discount is negotiated with Confluent sales and is not published; a person should confirm it.
    • 
    

  

  

    
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 a managed Kafka cluster in Confluent Cloud: the cluster type, the cloud, the networking, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the published cluster types and limits and returns a verdict, every reason and the capacity figures it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote a Standard Confluent Cloud Kafka cluster on AWS with 12 eCKU, private networking and a 99.99% uptime SLA. The workload peaks at 400 MBps of ingress and 1,100 MBps of egress, with 2,400 partitions and 11,000 client connections. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with six findings, all in one answer. The quote has 12 capacity units; this cluster type and cloud allows 1 to 10. The networking choice is not supported by this cluster type; Basic and Standard clusters use public networking, Enterprise and Freight clusters use private networking. The workload peaks at 400 MBps of ingress; the chosen cluster size supports 250 MBps. 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 limits that differ by cluster type and, for Dedicated, by cloud, a rules-based CPQ is error-prone for this.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the cluster type to enterprise and the stated ingress capacity to 720 MBps.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. All six earlier findings are cleared, and the validator's figure matches the quote.
    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.