Predictika · Industrial air filtration

An AI sales agent's Airpocket bag filter bank, checked before it goes out

An AI sales agent prepares a bank of Airpocket bag filters for an air handling unit: family, filter class and efficiency, size, pocket depth, number of pockets, quantity, frame, header depth and the airflow and pressure drop it states. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the performance tables MANN+HUMMEL publishes and returns a verdict, every reason and the figures it computed.

Scope: Airpocket EF, Airpocket Select Glass and Airpocket EF Glass fine-dust bag filters: published combinations, airflow, pressure drop, frame, heat and class. See everything this validator covers

This validator models banks of three MANN+HUMMEL bag filter families for HVAC air handling units, based on the performance tables in the MANN+HUMMEL Air Filtration Catalog, not just the scenarios shown below.

  • Published combinations — the ISO 16890 class, efficiency, size, pocket depth and number of pockets must be a row in the catalog's performance data for the family (23 rows for Airpocket EF, 14 for Select Glass, 26 for EF Glass). Each row gives the nominal airflow and initial pressure drop the validator uses.
  • Options — classes and pocket depths each family offers, the 20 or 25 mm header depth, the 70 degree Celsius heat limit, and the frame: glass fiber filters come with a metal frame only, while a plastic frame on an Airpocket EF is available on request and goes to a person.
  • Airflow and pressure — the bank's total rated airflow against the design airflow (Airpocket EF above its rating, or glass fiber filters more than 20 percent above or below it, go to a person), and the recommended final pressure drop (the lower of the initial value plus 100 Pa or three times the initial value, as the catalog states) against what the air handling unit allows.
  • Project requirement and stated figures — the filter class must meet the class the project requires, and the airflow and initial pressure drop the quote states must equal the catalog values.

Any bank of Airpocket EF, Select Glass or EF Glass filters that touches these rules is in scope — not only the scenarios pictured here. Not covered: prices (MANN+HUMMEL does not publish list prices for these filters, so this page shows no price build-up), Airpocket Access and Airpocket Select (synthetic), prefilter bag filters, panel, compact and HEPA filters, housings, and filter combinations the catalog does not list.

The AI sales agentBuilds the filter bank 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
Only the filter combinations the catalog publishes (64 rows across three families)Only with workarounds
Each family has its own table of class, efficiency, size, depth and pocket count. A rules-based CPQ needs a compatibility table or one rule per row, and every catalog update means editing them.
Class, efficiency, size, depth and pockets arrive in one request and are matched to the published rows on every call.
Final pressure drop that is the lower of two formulasOnly with care
The recommended final pressure drop is derived from the row's initial value by a two-way formula, then compared with the unit's limit; the formula has to be kept in step with the catalog.
The initial value comes from the matched row, the final value is computed, and both are checked on every call.
Glass fiber and synthetic filters with different optionsOnly with care
Frame options and the airflow range differ by family, so the same quote needs different rules depending on one earlier choice.
The family and each option are compared together, in any order.
Airflow and pressure drop the quote statesOnly with care
The stated figures are derived from the chosen row, so a rule has to recompute them and compare.
The stated figures are fields in the request, compared with the computed ones 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 selected filter class is not offered for this family; Airpocket EF is ePM10 and ePM1 only, while Select Glass and EF Glass are ePM10, ePM2.5 and ePM1.
2A pocket depth of 400 mm is not offered for the selected family; Airpocket EF offers 360, 500, 535, 600 and 635 mm, Select Glass 360, 500 and 635 mm, and EF Glass 360, 535 and 635 mm.
3The catalog lists no filter with 10 pockets, 85 percent efficiency and a depth of 600 mm in this size, class and family.
4The catalog lists no filter with 9 pockets, 85 percent efficiency and a depth of 635 mm in this size, class and family.
5The catalog lists no filter with 10 pockets, 85 percent efficiency and a depth of 635 mm in this size, class and family.
6A filter size of 400 by 400 mm is not a published size for these filters.
7A pocket depth of 535 mm is not offered for the selected family; Airpocket EF offers 360, 500, 535, 600 and 635 mm, Select Glass 360, 500 and 635 mm, and EF Glass 360, 535 and 635 mm.
8Glass fiber filters are made with a metal frame only; a plastic frame is not offered.
9A plastic frame for Airpocket EF is available only on request; a person should confirm it with MANN+HUMMEL.
10Header depth is 20 or 25 mm; the quote has 30 mm.
11The operating temperature of 71 degrees Celsius is above the 70 degree maximum heat resistance.
12The selected ISO 16,890 filter class is lower than the class the project requires.
13The recommended final pressure drop of 200 Pa is above the 199 Pa the air handling unit allows.
14The recommended final pressure drop of 105 Pa is above the 104 Pa the air handling unit allows.
15The quote states 40,000 cubic meters an hour of nominal airflow; the selected filters give 40800.
16The quote states an initial pressure drop of 70 Pa; the catalog lists 75 Pa for the selected filter.

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. Class, heat and approvals: 6 Airpocket EF filters on a hot duct, class ePM10, with a plastic frame

The agent's request. Please quote 6 Airpocket EF bag filters, 592 by 592 by 635 mm with 6 pockets, ISO 16890 class ePM10 at 70 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 21,600 cubic meters an hour. The unit allows a final pressure drop of 250 Pa and the air is at 80 degrees Celsius. The filters have a plastic frame and a 25 mm header. The quote states 20,400 cubic meters an hour of nominal airflow and an initial pressure drop of 40 Pa. It is ready to send.

Verdict: Blocked.

  • Operating temperature above the limit. The operating temperature of 80 degrees Celsius is above the 70 degree maximum heat resistance.
  • Filter class below the project requirement. The selected ISO 16,890 filter class is lower than the class the project requires.
  • Plastic frame is available on request. A plastic frame for Airpocket EF is available only on request; a person should confirm it with MANN+HUMMEL.
  • Design airflow above the nominal airflow. The design airflow of 21,600 cubic meters an hour is above the 20,400 the selected filters are rated for in total; a person should confirm the pressure drop at that airflow.

2. Class, heat and approvals: The agent resubmits after fixing the errors

The agent's request. Please quote 6 Airpocket EF bag filters, 592 by 592 by 635 mm with 6 pockets, ISO 16890 class ePM1 at 65 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 21,600 cubic meters an hour. The unit allows a final pressure drop of 250 Pa and the air is at 60 degrees Celsius. The filters have a plastic frame and a 25 mm header. The quote states 20,400 cubic meters an hour of nominal airflow and an initial pressure drop of 100 Pa. It is ready to send.

Verdict: Needs approval.

  • Plastic frame is available on request. A plastic frame for Airpocket EF is available only on request; a person should confirm it with MANN+HUMMEL.
  • Design airflow above the nominal airflow. The design airflow of 21,600 cubic meters an hour is above the 20,400 the selected filters are rated for in total; a person should confirm the pressure drop at that airflow.
  • Filter bank summary. 6 filters are rated for 20,400 cubic meters an hour in total, with an initial pressure drop of 100 Pa and a recommended final pressure drop of 200 Pa.

3. Class, heat and approvals: The agent resubmits after fixing the errors

The agent's request. Please quote 7 Airpocket EF bag filters, 592 by 592 by 635 mm with 6 pockets, ISO 16890 class ePM1 at 65 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 21,600 cubic meters an hour. The unit allows a final pressure drop of 250 Pa and the air is at 60 degrees Celsius. The filters have a metal frame and a 25 mm header. The quote states 23,800 cubic meters an hour of nominal airflow and an initial pressure drop of 100 Pa. It is ready to send.

Verdict: Allowed.

  • Filter bank summary. 7 filters are rated for 23,800 cubic meters an hour in total, with an initial pressure drop of 100 Pa and a recommended final pressure drop of 200 Pa.

4. Pressure drop and stated figures: A bank of 12 Airpocket EF filters for a hospital unit

The agent's request. Please quote 12 Airpocket EF bag filters, 592 by 592 by 635 mm with 8 pockets, ISO 16890 class ePM1 at 85 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 40,800 cubic meters an hour. The unit allows a final pressure drop of 180 Pa and the air is at 40 degrees Celsius. The filters have a metal frame and a 25 mm header. The quote states 40,800 cubic meters an hour of nominal airflow and an initial pressure drop of 90 Pa. It is ready to send.

Verdict: Blocked.

  • Final pressure drop above the system limit. The recommended final pressure drop of 200 Pa is above the 180 Pa the air handling unit allows.
  • Stated pressure drop does not match. The quote states an initial pressure drop of 90 Pa; the catalog lists 100 Pa for the selected filter.

5. Pressure drop and stated figures: The agent resubmits after fixing the errors

The agent's request. Please quote 12 Airpocket EF bag filters, 592 by 592 by 635 mm with 10 pockets, ISO 16890 class ePM1 at 85 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 40,800 cubic meters an hour. The unit allows a final pressure drop of 180 Pa and the air is at 40 degrees Celsius. The filters have a metal frame and a 25 mm header. The quote states 40,800 cubic meters an hour of nominal airflow and an initial pressure drop of 75 Pa. It is ready to send.

Verdict: Allowed.

  • Filter bank summary. 12 filters are rated for 40,800 cubic meters an hour in total, with an initial pressure drop of 75 Pa and a recommended final pressure drop of 175 Pa.

6. Frame and header on glass fiber filters: 8 Airpocket Select Glass filters with a plastic frame and a 30 mm header

The agent's request. Please quote 8 Airpocket Select Glass bag filters, 592 by 592 by 635 mm with 10 pockets, ISO 16890 class ePM1 at 85 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 27,200 cubic meters an hour. The unit allows a final pressure drop of 300 Pa and the air is at 40 degrees Celsius. The filters have a plastic frame and a 30 mm header. The quote states 27,200 cubic meters an hour of nominal airflow and an initial pressure drop of 170 Pa. It is ready to send.

Verdict: Blocked.

  • Header depth not available. Header depth is 20 or 25 mm; the quote has 30 mm.
  • Glass fiber filters come with a metal frame only. Glass fiber filters are made with a metal frame only; a plastic frame is not offered.

7. Frame and header on glass fiber filters: The agent resubmits after fixing the errors

The agent's request. Please quote 8 Airpocket Select Glass bag filters, 592 by 592 by 635 mm with 10 pockets, ISO 16890 class ePM1 at 85 percent efficiency, for a hospital air handling unit that needs class ePM1 filtration and moves 27,200 cubic meters an hour. The unit allows a final pressure drop of 300 Pa and the air is at 40 degrees Celsius. The filters have a metal frame and a 25 mm header. The quote states 27,200 cubic meters an hour of nominal airflow and an initial pressure drop of 170 Pa. It is ready to send.

Verdict: Allowed.

  • Filter bank summary. 8 filters are rated for 27,200 cubic meters an hour in total, with an initial pressure drop of 170 Pa and a recommended final pressure drop of 270 Pa.

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 bank of Airpocket bag filters for an air handling unit: family, filter class and efficiency, size, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the performance tables MANN+HUMMEL publishes and returns a verdict, every reason and the figures it computed.
  2. I will start with the first request by the agent.
  3. The agent says. Please quote 6 Airpocket EF bag filters, 592 by 592 by 635 mm with 6 pockets, ISO 16890 class ePM10 at 70 percent efficiency. The unit allows a final pressure drop of 250 Pa and the air is at 80 degrees Celsius. The rest of the request is on screen. The quote states 20,400 cubic meters an hour of nominal airflow and an initial pressure drop of 40 Pa.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with two findings and two approvals, all in one answer. The operating temperature of 80 degrees Celsius is above the 70 degree maximum heat resistance. The selected ISO 16,890 filter class is lower than the class the project requires.
  6. The validator returned two 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. What this shows: for only the filter combinations the catalog publishes (64 rows across three families), a rules-based CPQ can do this only with workarounds.
  8. Now the agent resubmits.
  9. The corrected request. It changes the ISO 16890 class to ePM1, the efficiency to 65 percent, the air temperature to 60 degrees Celsius and the stated initial pressure drop as to 100 Pa.
  10. It calls the validator again.
  11. Not blocked, but held for a person.
  12. The two blocking findings are cleared; two approvals remain for a person: plastic frame is available on request and design airflow above the nominal airflow.
  13. The agent resubmits once more.
  14. The revised request. It changes the quantity to 7, the frame to metal and the stated nominal airflow as to 23,800 cubic meters an hour.
  15. It calls the validator once more.
  16. This time the validator clears it.
  17. Both earlier findings are cleared, and the validator's figure matches the quote.
  18. Four 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.