Predictika · Application delivery orders

An AI sales agent's F5 BIG-IP rSeries order, checked before it goes out

An AI sales agent prepares an F5 BIG-IP rSeries order: the appliance model, the BIG-IP tenants that will run on it, a Good, Better or Best software bundle with module add-ons, optics, power, tenant deployment and support. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the rSeries platform limits, tenant sizing, bundles, redundancy, lead time and discount, and returns a verdict, every reason, a line-item quote and the figures it computed.

Scope: F5 BIG-IP rSeries orders: the r2600 and r2800, r4600 and r4800, r5600, r5800, r5900 and r5920-DF, r10600, r10800, r10900 and r10920-DF, and r12600-DS, r12800-DS and r12900-DS appliances, with tenants, Good / Better / Best bundles, DNS, AFM, Advanced WAF and APM add-ons, optics, DC power and support. See everything this validator covers

This validator models F5 BIG-IP rSeries orders using the platform facts in F5's public rSeries Planning Guide, not just the scenarios shown below.

  • Platform limits (from the guide) — the vCPUs each model gives to tenants, tenant sizes of 1 vCPU or an even number, the minimum tenant memory for its vCPUs and the memory a model has for tenants, the 100G / 40G and 25G / 10G ports of each model, dual disks only on some models, the integrated FIPS HSM only on the r5920-DF and r10920-DF, and tenant-level HA only between separate appliances.
  • Bundles and modules (modeled) — what the Good, Better and Best bundles include, add-ons that duplicate a bundle, and the minimum tenant size for Advanced WAF and Access Policy Manager.
  • Terms — DC power, Premier support with an HA pair, support terms of 1, 3 or 5 years, lead time against the date the customer needs the appliances, discount and the stated total.
  • Approvals — tenants that use every vCPU of the model, a discount above 10 percent (up to 20 percent), a five-year term and a total above $1.5 million go to a person; a discount above 20 percent is blocked.

Any BIG-IP rSeries order that touches these models, tenants, bundles and options is in scope — not only the scenarios pictured here. Not covered: VELOS, iSeries, BIG-IP Virtual Edition, VIPRION and other F5 products, taxes and freight. The model names and platform limits come from F5's public guide; the bundle contents, module rules, lead times and every price on this page are modeled or illustrative.

The AI sales agentBuilds the rSeries order 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
Tenant sizing against the model limitsOnly with workarounds
Tenants must fit the vCPUs the model gives them, the memory it has for tenants and the minimum memory for each tenant size, and the formulas differ between model families. A CPQ holds tenants as free text or not at all, so the arithmetic is done by an engineer.
Tenants, vCPUs and memory are inputs, and every limit of the chosen model is checked on every call.
Features that only some models haveOnly with care
FIPS key storage, dual disks, DC power and 100G ports exist on some models only. A rule per feature is possible, but someone has to keep the list of models current.
The customer's needs are compared with what the chosen model has.
Bundles and add-ons that overlapOnly with care
An add-on for a module that the bundle already includes is a quiet double charge. It takes a rule per module and bundle to catch it.
Each add-on is checked against the contents of the chosen bundle.
Module needs against tenant sizeOnly with workarounds
Advanced WAF and Access Policy Manager need tenants of a minimum size, and the bundle on one line and the tenant size on another are rarely tied together.
The bundle, the add-ons and the tenant size are checked against each other.
HA that has to be between appliancesOnly with workarounds
rSeries HA is between tenants on separate appliances, so the appliance count has to be a pair. A flat quote lets one appliance carry an HA design.
The HA design and the appliance count are checked together.
A stated total that has to agree with the quoteOnly with workarounds
Appliances, bundles, add-ons, optics, services and years of support make a total that can drift from what is typed on the quote without anyone noticing.
The total is computed from the choices and compared with the total the quote states.
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 no rSeries appliance.
2The quote has 1 appliances in a tenant-level HA design; rSeries HA runs between tenants on separate appliances, so it takes pairs of 2, 4, 6 or 8.
3The quote has no BIG-IP tenant; rSeries runs only tenants, with no bare-metal option.
4The quote has tenants of 3 vCPUs; tenants come in 1 vCPU or an even number of vCPUs.
5The tenants use all 28 vCPUs the model gives them, leaving no headroom, so a solution engineer should confirm.
6The tenants need 32 vCPUs on each appliance, and this model gives tenants 28.
7A tenant with 28 GB of memory is below the recommended minimum of 29,184 MB for its vCPUs.
8The tenants need 234 GB of memory on each appliance, and this model has 231 GB for tenants.
9The quote uses 5 100G / 40G ports on each appliance, and this model has 4.
10The quote uses 17 25G / 10G ports on each appliance, and this model has 16.
11The customer needs FIPS key storage, and only the r5,920-DF and r10,920-DF have the integrated HSM.
12The customer needs redundant disks, and this model has a single disk; dual disks come on the r5,920-DF, r10,000 and r12,000.
13The quote has DC power on an r2000, r4,000 or r5,000 model; DC power is quoted on the r10,000 and r12,000 only.
14The quote adds BIG-IP DNS to a bundle that already includes it.
15The quote adds Advanced Firewall Manager to a bundle that already includes it.
16The quote adds Advanced WAF to the Best bundle, which already includes it.
17The quote adds Access Policy Manager to the Best bundle, which already includes it.
18The quote runs Advanced WAF on tenants of 2 vCPUs; it needs at least 4.
19The quote runs Access Policy Manager on tenants of 2 vCPUs; it needs at least 4.
20The quote has Premier support on a standalone design; Premier support is quoted with a tenant-level HA pair.
21The quote has 5 years of support, a long price commitment, so a person should confirm.
22The quote has 2 years of support; support is quoted for 1, 3 or 5 years.
23The customer needs the appliances in 3 weeks, and the lead time for this order is 4 weeks.
24The quote takes a discount of $54,831 against $548,300 of list price, above the 10 percent that needs a sales director.

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. Retailer data-center refresh: An order for a retailer data-center refresh: twelve tenants with FIPS key storage, dual disks and DC power on an r5900

The agent's request. Please quote 3 F5 BIG-IP r5900 appliances in a tenant-level HA design with the Best bundle. It adds Advanced WAF. Each appliance runs 12 tenants of 4 vCPUs and 8 GB of memory each. The customer needs FIPS key storage and redundant disks. Each appliance uses 2 100G / 40G and 8 25G / 10G ports and DC power. Support is Premier for 3 years. The customer needs the appliances in 6 weeks. The quote states $1,766,000. It is ready to send.

Verdict: Blocked.

  • Tenant HA needs pairs of appliances. The quote has 3 appliances in a tenant-level HA design; rSeries HA runs between tenants on separate appliances, so it takes pairs of 2, 4, 6 or 8.
  • Tenant vCPUs over the model limit. The tenants need 48 vCPUs on each appliance, and this model gives tenants 26.
  • Tenant memory under the minimum. A tenant with 8 GB of memory is below the recommended minimum of 14,848 MB for its vCPUs.
  • FIPS needed on a model without the HSM. The customer needs FIPS key storage, and only the r5,920-DF and r10,920-DF have the integrated HSM.
  • Redundant disk on a single-disk model. The customer needs redundant disks, and this model has a single disk; dual disks come on the r5,920-DF, r10,000 and r12,000.
  • DC power on a model without it. The quote has DC power on an r2000, r4,000 or r5,000 model; DC power is quoted on the r10,000 and r12,000 only.
  • Advanced WAF add-on already in the bundle. The quote adds Advanced WAF to the Best bundle, which already includes it.
  • Stated total does not match. The quote states a total of $1,766,000; the prices give $806,700.

2. Retailer data-center refresh: The agent resubmits after fixing the errors

The agent's request. Please quote 4 F5 BIG-IP r10920-DF appliances in a tenant-level HA design with the Best bundle. Each appliance runs 9 tenants of 4 vCPUs and 15 GB of memory each. The customer needs FIPS key storage and redundant disks. Each appliance uses 4 100G / 40G and 8 25G / 10G ports and DC power. Support is Premier for 3 years. The customer needs the appliances in 12 weeks. The discount is $212,712. The quote states $1,559,888. It is ready to send.

Verdict: Needs approval.

  • Discount above 10 percent. The quote takes a discount of $212,712 against $1,772,600 of list price, above the 10 percent that needs a sales director.
  • Tenants use every vCPU. The tenants use all 36 vCPUs the model gives them, leaving no headroom, so a solution engineer should confirm.
  • Total over $1.5 million. The total is $1,559,888, above $1.5 million, so a person should confirm.
  • Quote summary. List price of $1,772,600, discount of $212,712 and a total of $1,559,888.

3. Retailer data-center refresh: The agent resubmits after fixing the errors

The agent's request. Please quote 2 F5 BIG-IP r10920-DF appliances in a tenant-level HA design with the Best bundle. Each appliance runs 8 tenants of 4 vCPUs and 15 GB of memory each. The customer needs FIPS key storage and redundant disks. Each appliance uses 4 100G / 40G and 8 25G / 10G ports and DC power. Support is Premier for 3 years. The customer needs the appliances in 12 weeks. The discount is $44,250. The quote states $840,750. It is ready to send.

Verdict: Allowed.

  • Quote summary. List price of $885,000, discount of $44,250 and a total of $840,750.

4. E-commerce platform: An order for an e-commerce platform team: an r4800 pair with 100G ports and DNS added to the Better bundle

The agent's request. Please quote 2 F5 BIG-IP r4800 appliances in a tenant-level HA design with the Better bundle. It adds BIG-IP DNS. Each appliance runs 2 tenants of 8 vCPUs and 24 GB of memory each. Each appliance uses 2 100G / 40G and 4 25G / 10G ports. Support is Premium for 3 years. The customer needs the appliances in 10 weeks. The discount is $6,752. The quote states $162,048. It is ready to send.

Verdict: Blocked.

  • More 100G / 40G ports than the model has. The quote uses 2 100G / 40G ports on each appliance, and this model has 0.
  • DNS add-on already in the bundle. The quote adds BIG-IP DNS to a bundle that already includes it.
  • Tenants use every vCPU. The tenants use all 16 vCPUs the model gives them, leaving no headroom, so a solution engineer should confirm.

5. E-commerce platform: The agent resubmits after fixing the errors

The agent's request. Please quote 2 F5 BIG-IP r5800 appliances in a tenant-level HA design with the Better bundle. Each appliance runs 2 tenants of 8 vCPUs and 30 GB of memory each. Each appliance uses 2 100G / 40G and 6 25G / 10G ports. Support is Premium for 3 years. The customer needs the appliances in 10 weeks. The discount is $12,176. The quote states $292,224. It is ready to send.

Verdict: Allowed.

  • Quote summary. List price of $304,400, discount of $12,176 and a total of $292,224.

6. University lab: An order for a university lab: the Best bundle on small tenants with Premier support on one appliance

The agent's request. Please quote 1 F5 BIG-IP r2800 appliance with the Best bundle. Each appliance runs 3 tenants of 2 vCPUs and 6 GB of memory each. Each appliance uses 0 100G / 40G and 4 25G / 10G ports. Support is Premier for 2 years. The customer needs the appliances in 6 weeks. The quote states $51,250. It is ready to send.

Verdict: Blocked.

  • Advanced WAF on tenants that are too small. The quote runs Advanced WAF on tenants of 2 vCPUs; it needs at least 4.
  • APM on tenants that are too small. The quote runs Access Policy Manager on tenants of 2 vCPUs; it needs at least 4.
  • Premier support without HA. The quote has Premier support on a standalone design; Premier support is quoted with a tenant-level HA pair.
  • Support term not offered. The quote has 2 years of support; support is quoted for 1, 3 or 5 years.
  • Stated total does not match. The quote states a total of $51,250; the prices give $81,250.

7. University lab: The agent resubmits after fixing the errors

The agent's request. Please quote 1 F5 BIG-IP r2800 appliance with the Good bundle. Each appliance runs 3 tenants of 2 vCPUs and 6 GB of memory each. Each appliance uses 0 100G / 40G and 4 25G / 10G ports. Support is Standard for 3 years. The customer needs the appliances in 6 weeks. The quote states $43,850. It is ready to send.

Verdict: Allowed.

  • Quote summary. List price of $43,850 and a total of $43,850.

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 an F5 BIG-IP rSeries order: the appliance model, the BIG-IP tenants that will run on it, a Good, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the rSeries platform limits, tenant sizing, bundles, redundancy, lead time and discount, and returns a verdict, every reason, a line-item quote and the figures it computed.
  2. I will start with the first request by the agent.
  3. The agent says. Please quote 3 F5 BIG-IP r5900 appliances in a tenant-level HA design with the Best bundle. It adds Advanced WAF. The rest of the request is on screen. The quote states $1,766,000.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with eight findings, all in one answer. The quote has 3 appliances in a tenant-level HA design; rSeries HA runs between tenants on separate appliances, so it takes pairs of 2, 4, 6 or 8. The tenants need 48 vCPUs on each appliance, and this model gives tenants 26. A tenant with 8 GB of memory is below the recommended minimum of 14,848 MB for its vCPUs. The other five are on screen.
  6. The validator returned eight blocking findings in one answer, led by Tenant HA needs pairs of appliances and Tenant vCPUs over the model limit, among others. A check that reads one product or one field at a time would pass most of this; it shows only when model limits, tenants, memory, bundles, options and the stated total are checked against each other.
  7. What this shows: for tenant sizing against the model limits, a rules-based CPQ can do this only with workarounds.
  8. Now the agent resubmits.
  9. The corrected request. It changes the appliance model to BIG-IP r10920-DF, the appliances to 4, the Advanced WAF add-on to no, the tenants per appliance to 9 and the stated total to $1,559,888. Four more changes are on screen.
  10. It calls the validator again.
  11. Not blocked, but held for a person.
  12. The validator also returns the line-item quote. It starts with BIG-IP rSeries appliance and BIG-IP software bundle; the other five lines are on screen.
  13. The eight blocking findings are cleared; three approvals remain for a person: discount above 10 percent, Tenants use every vCPU and total over $1.5 million.
  14. The agent resubmits once more.
  15. The revised request. It changes the appliances to 2, the tenants per appliance to 8, the discount to $44,250 and the stated total to $840,750.
  16. It calls the validator once more.
  17. This time the validator clears it.
  18. The validator also returns the line-item quote. It starts with BIG-IP rSeries appliance and BIG-IP software bundle; the other five lines are on screen.
  19. All three earlier findings are cleared, and the validator's total, $840,750, matches the quote.
  20. Eight problems caught in one call, fixed in two passes, and a clean $840,750 quote. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.