Predictika · Building systems service agreements

An AI sales agent's building service agreement, checked before it goes out

An AI sales agent prepares a service and maintenance agreement for a building owner: emergency service, preventative maintenance, hourly services, rental equipment, the contract basis and the term. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the service line and returns a verdict, every reason and the figures it computed.

Scope: Limbach Owner Direct Relationships service and maintenance agreements: 24/7 emergency service, preventative maintenance, asset valuation, staff augmentation, building automation consultation, hydro-jetting, training and rental equipment. See everything this validator covers

This validator models service and maintenance agreement quotes in the style of Limbach's Owner Direct Relationships (ODR) business, not just the scenarios shown below.

  • Services — the predictive maintenance services Limbach lists (24/7 emergency service and repairs, on-site asset valuation, staff augmentation, proactive preventative maintenance, building automation consultation, hydro-jetting and facility team training) and turnkey rental equipment; contract bases fixed-price, modified fixed-price and time and materials, with evergreen contracts.
  • Facility rules — healthcare, data center and life sciences facilities must include 24/7 emergency service and at least four preventative maintenance visits a year; maintenance and valuation are priced per asset; hourly services cannot be fixed price; rental equipment needs a compatibility survey*.
  • Term and total — terms of one to five years (over three years go to a person), time and materials on an evergreen contract, large annual values and long rentals go to a person, and the total the quote states is compared with the total the typical rates give. Items marked with an asterisk (*) are illustrative: the separately priced compatibility survey.

Any agreement for these services that touches these rules is in scope — not only the scenarios pictured here. Not covered: Limbach's General Contractor Relationships (new construction bids), design-build projects, replacements and retrofits, infrastructure upgrades and energy efficiency projects. Limbach does not publish rates or rules: every rule and price here is typical for this kind of service, not published by Limbach; all customers and agreements are fictional.

The AI sales agentBuilds the service agreement 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
Requirements that depend on the kind of facilityOnly with care
Emergency service and visit frequency depend on the facility type chosen on another line; each type needs its own rule and every new facility type is more rules to maintain.
The facility type and every service are in one request and the requirements are checked together.
Services that cannot go with a contract basisOnly with workarounds
Hourly services on a fixed-price contract is a cross-line conflict that depends on the basis and on hours entered elsewhere, and it interacts with the other checks.
Basis and hours are checked together on every call, whatever order they arrive in.
Rental equipment that needs a survey firstOnly with care
A dependency between a rental line and a survey line; a rule can enforce it, but it has to be written and kept in step.
Rental weeks and the survey are in one request and the dependency is checked with everything else.
Term, evergreen and value thresholds that send a quote to a personOnly with workarounds
Approval thresholds depend on the computed annual value and on the term, so they usually live in an approval process apart from the quote rules and are easy to miss.
Thresholds are evaluated on the computed values and returned as findings with the rest.
Stated total that must match the ratesOnly with care
A stated figure is a free-entry field; a rule can compare it with the computed total, but someone has to write and maintain it.
The stated total travels with the quote and is compared with the computed total on every call.

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.

1Healthcare, data center and life sciences facilities must include 24/7 emergency service and repairs; the quote has none.
2A mission-critical facility needs at least 4 preventative maintenance visits a year; the quote has 3.
3Preventative maintenance and asset valuation are priced per asset and the quote lists 0 assets.
4Rental equipment includes system design and checking compatibility with the existing system; the quote has 8 rental weeks and no compatibility survey.
5Staff augmentation and building automation consultation are billed by the hour; they cannot be on a fixed-price basis.
6An evergreen contract billed on time and materials has no ceiling; a person should confirm a not-to-exceed amount.
7The agreement term is 4 years; terms over 3 years need a person to confirm the price escalation.
8The agreement term is 5 years; terms over 3 years need a person to confirm the price escalation.
9The agreement term is 6 years; this model quotes terms of 1 to 5 years.
10The agreement term is 0 years; this model quotes terms of 1 to 5 years.
11The quote states an annual total of 15,000 dollars; the typical rates give 37,200 dollars.
12The annual total is 251,400 dollars; agreements over 250,000 dollars a year need a person to confirm.
13The rental runs 27 weeks; a person should consider a replacement instead of a long rental.

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. Hospital campus: A hospital agreement on a fixed price with no emergency service and two visits a year

The agent's request. Please quote a 1-year fixed-price service agreement (evergreen) for a healthcare facility with 320 assets and 2 preventative maintenance visits a year, including 300 hours of staff augmentation. The quote states an annual total of 104,400 dollars. It is ready to send.

Verdict: Blocked.

  • Mission-critical facility needs 24/7 emergency service. Healthcare, data center and life sciences facilities must include 24/7 emergency service and repairs; the quote has none.
  • Preventative maintenance too infrequent. A mission-critical facility needs at least 4 preventative maintenance visits a year; the quote has 2.
  • Hourly services cannot be fixed price. Staff augmentation and building automation consultation are billed by the hour; they cannot be on a fixed-price basis.

2. Hospital campus: The agent resubmits after fixing the errors

The agent's request. Please quote a 3-year modified fixed-price service agreement (evergreen) for a healthcare facility with 320 assets and 4 preventative maintenance visits a year, including 24/7 emergency service and 300 hours of staff augmentation. The quote states an annual total of 199,200 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. 320 assets, 4 preventative maintenance visits a year, annual total 199,200 dollars at typical rates for a 3 year agreement.

3. Data center with rental equipment: A data center agreement with rental equipment, a six-year term and a wrong stated total

The agent's request. Please quote a 6-year modified fixed-price service agreement (evergreen) for a data center with 180 assets and 4 preventative maintenance visits a year, including 24/7 emergency service and 10 weeks of rental equipment. The quote states an annual total of 170,000 dollars. It is ready to send.

Verdict: Blocked.

  • Rental equipment needs a compatibility survey. Rental equipment includes system design and checking compatibility with the existing system; the quote has 10 rental weeks and no compatibility survey.
  • Term outside what can be quoted. The agreement term is 6 years; this model quotes terms of 1 to 5 years.
  • Stated total does not match. The quote states an annual total of 170,000 dollars; the typical rates give 152,400 dollars.

4. Data center with rental equipment: The agent resubmits after fixing the errors

The agent's request. Please quote a 4-year modified fixed-price service agreement (evergreen) for a data center with 180 assets and 4 preventative maintenance visits a year, including 24/7 emergency service, 10 weeks of rental equipment and a compatibility survey*. The quote states an annual total of 155,600 dollars. It is ready to send.

Verdict: Needs approval.

  • Long price commitment to confirm. The agreement term is 4 years; terms over 3 years need a person to confirm the price escalation.
  • Order summary. 180 assets, 4 preventative maintenance visits a year, annual total 155,600 dollars at typical rates for a 4 year agreement.

5. Data center with rental equipment: The agent resubmits after fixing the errors

The agent's request. Please quote a 3-year modified fixed-price service agreement (evergreen) for a data center with 180 assets and 4 preventative maintenance visits a year, including 24/7 emergency service, 10 weeks of rental equipment and a compatibility survey*. The quote states an annual total of 155,600 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. 180 assets, 4 preventative maintenance visits a year, annual total 155,600 dollars at typical rates for a 3 year agreement.

6. University agreement: A university time-and-materials agreement with no assets listed

The agent's request. Please quote a 1-year time and materials service agreement (evergreen) for a higher education campus with 0 assets and 4 preventative maintenance visits a year, including 24/7 emergency service, 60 hours of building automation consultation and 2 hydro-jetting visits. The quote states an annual total of 30,800 dollars. It is ready to send.

Verdict: Blocked.

  • Services priced per asset but no assets. Preventative maintenance and asset valuation are priced per asset and the quote lists 0 assets.
  • Time and materials on an evergreen contract. An evergreen contract billed on time and materials has no ceiling; a person should confirm a not-to-exceed amount.

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

The agent's request. Please quote a 1-year modified fixed-price service agreement (evergreen) for a higher education campus with 150 assets and 4 preventative maintenance visits a year, including 24/7 emergency service, 60 hours of building automation consultation and 2 hydro-jetting visits. The quote states an annual total of 102,800 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. 150 assets, 4 preventative maintenance visits a year, annual total 102,800 dollars at typical rates for a 1 year agreement.

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 service and maintenance agreement for a building owner: emergency service, preventative maintenance, hourly services, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the service line 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 a 6-year modified fixed-price service agreement (evergreen) for a data center. The quote states an annual total of 170,000 dollars. The rest of the request is on screen.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with three findings, all in one answer. Rental equipment includes system design and checking compatibility with the existing system; the quote has 10 rental weeks and no compatibility survey. The agreement term is 6 years; this model quotes terms of 1 to 5 years. The quote states an annual total of 170,000 dollars; the typical rates give 152,400 dollars.
  6. The validator returned three 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. What this shows: for rental equipment that needs a survey first, a rules-based CPQ is error-prone for this.
  8. Now the agent resubmits.
  9. The corrected request. It changes the term to 4 years, the compatibility survey* to yes and the stated annual total to 155,600 dollars.
  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 Preventative maintenance and 24/7 emergency service retainer; the other two lines are on screen.
  13. The three blocking findings are cleared; one approval remains for a person: long price commitment to confirm.
  14. The agent resubmits once more.
  15. The revised request. It changes the term to 3 years.
  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 Preventative maintenance and 24/7 emergency service retainer; the other two lines are on screen.
  19. The earlier finding is cleared, and the validator's total, $155,600, matches the quote.
  20. Three 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. Now you can try the other scenarios yourself.