Predictika · Clinical decision support subscriptions

An AI sales agent's Micromedex order, checked before it goes out

An AI sales agent prepares a subscription order for clinical decision support: a Micromedex or DynaMedex package for a health system, AI search add-ons, integrated components for the hospital information system and a RED BOOK data feed. Before the order goes out, the agent sends it to Predictika, which checks it against a declarative model, compares the bed count the agent states with the one it computes, and returns a verdict, every reason and the figures it computed.

Scope: Micromedex for HealthCare Centers and Micromedex for Academic: Clinical Evidence packages, DynaMedex, AI add-ons, POISINDEX and NeoFax components, integrated components and the RED BOOK Local Feature. See everything this validator covers

This validator models Micromedex subscription orders built from the product names and terms in Merative's public Micromedex Service Description, not just the scenarios shown below.

  • Packages and segments — the Clinical Evidence Standard, Plus and Premium Packages and the DynaMedex Package for health care centers, and the Academic Package for classroom or preceptor use; the academic packages cannot be used in place of productive hospital use, and a teaching order has no integrated components.
  • AI add-ons — Micromedex AI-Powered Search and DynaMed's Dyna AI (an add-on for DynaMedex only), both English only; DynaMedex cannot be combined with a separate Micromedex package.
  • Components and integration — the separate POISINDEX and NeoFax and Pediatric Drug components against what each package already includes, and CareNotes Integrated, the Infobutton and the embedded API against an implementation service.
  • RED BOOK Local Feature — the Advanced, Expanded, Select and J-Code packages and their update frequency, the need for health information system integration, and the rule that RED BOOK is a look-up tool and not for automated claims processing.
  • Size and term — staffed beds across facilities (compared with the figure the order states), students and faculty for academic orders, satellite locations and a term of 1 to 5 years, with reviews for large health systems.

Any Micromedex or DynaMedex subscription order that touches these products and rules is in scope, not only the scenarios pictured here. Product names, charge metrics, the English-only AI terms, the DynaMedex and academic limits and the RED BOOK package list are from the public Service Description. What each package contains beyond its description, the size and term thresholds, the review rules and the implementation requirement are modeled from comparable software and are not Merative-published rules. Not covered: prices (this validator produces no price build-up or line-item quote), OrbitalRX, SmartEngage, the pharmaceutical manufacturer offerings and other Merative products.

The AI sales agentBuilds the subscription order 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
Stated bed count that must equal facilities times bedsOnly with care
A stated figure is a free-entry field; a rule can compare it with a computed total, but someone has to write and maintain that comparison.
The stated bed count travels with the order and is compared with the computed total on every call.
An AI add-on that works with one package only, and only in EnglishOnly with workarounds
The add-on depends on the package on one line and the language on another, so it needs cross-line rules that interact with the rest.
Package, add-on and language are in one request and checked together.
A component already included in a higher packageOnly with care
A rule per package and component pair; it works, but every new package or component means more rules to maintain.
Included content is looked up from the package, so duplicates are caught for every combination.
Integrated components that need an implementation serviceOnly with workarounds
The dependency crosses several lines, and a missing service is usually found by a person at the end.
Every component and service is in one request, so the gap is found whichever order the choices arrive in.
Data feed package and update frequency that are not offered together, and a use that the terms forbidOnly with workarounds
A list of allowed pairs and a free-text intended use; the intended use is rarely a field at all.
The package, frequency and intended use are checked together, with the reason stated.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the order 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 order; approval findings need a person to sign off; everything else is computed.

1A hospital order needs staffed beds and an academic order needs the number of students and faculty; the order has neither.
2The Academic packages are for teaching use only and cannot replace or supplement Micromedex offerings for productive use in a hospital.
3A teaching institution is ordering a Healthcare Centers package; a person should confirm it will be used in clinical practice.
4CareNotes Integrated, the Infobutton and the embedded API are for integration into a health information system; a teaching order covers classroom use only.
5Dyna AI is an add-on that works with DynaMedex only; the order does not have the DynaMedex Package.
6Micromedex AI-Powered Search and Dyna AI are designed and supported for English only; the order asks for another language.
7When the DynaMedex Package is provided, DynaMed and Micromedex cannot be used separately from DynaMedex; the order also has a separate Micromedex package.
8The Clinical Evidence Premium Package already includes in-depth toxicology with POISINDEX; the separate POISINDEX component duplicates it.
9This package already includes NeoFax and Pediatrics; the separate NeoFax and Pediatric Drug component duplicates it.
10An integrated component (CareNotes Integrated, the Infobutton or the embedded API) is set up in the hospital information system; the order has no implementation service.
11The RED BOOK package and update frequency are not offered together; Advanced is offered weekly or monthly and the J-Code packages quarterly.
12The RED BOOK Local Feature delivers database files for integration into the hospital health information system; the order has no such integration.
13RED BOOK may be used only as a referential look-up tool and not for an automated claims processing system.
14The order states 700 staffed beds in total, but the facilities on it give 600.
15An order for 2,001 staffed beds needs an enterprise account review before it goes out.
16A term of 5 years needs a commercial review of price protection.
17A subscription term runs 1 to 5 years in this model; the order has 6 years.
1826 non-acute satellite locations need a review of the location list.

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. Regional health system: A six-hospital system orders Premium with Dyna AI, Spanish AI Search, RED BOOK for claims and CareNotes Integrated

The agent's request. Please prepare a 3-year order for the Micromedex Clinical Evidence Premium Package for 6 hospitals with 400 staffed beds each and 18 non-acute satellite locations, with Dyna AI and Micromedex AI-Powered Search in Spanish, the separate POISINDEX component, CareNotes Integrated and the RED BOOK Local Feature Advanced package updated quarterly for the automated claims processing system. The order states 2,000 staffed beds in total. It is ready to send.

Verdict: Blocked.

  • Dyna AI needs DynaMedex. Dyna AI is an add-on that works with DynaMedex only; the order does not have the DynaMedex Package.
  • AI features are English only. Micromedex AI-Powered Search and Dyna AI are designed and supported for English only; the order asks for another language.
  • POISINDEX already in the package. The Clinical Evidence Premium Package already includes in-depth toxicology with POISINDEX; the separate POISINDEX component duplicates it.
  • Integrated components need implementation. An integrated component (CareNotes Integrated, the Infobutton or the embedded API) is set up in the hospital information system; the order has no implementation service.
  • RED BOOK package and frequency not offered together. The RED BOOK package and update frequency are not offered together; Advanced is offered weekly or monthly and the J-Code packages quarterly.
  • RED BOOK for claims processing. RED BOOK may be used only as a referential look-up tool and not for an automated claims processing system.
  • Stated beds differ from the order. The order states 2,000 staffed beds in total, but the facilities on it give 2,400.
  • Large health system to review. An order for 2,400 staffed beds needs an enterprise account review before it goes out.

2. Regional health system: The agent resubmits after fixing the errors

The agent's request. Please prepare a 3-year order for the DynaMedex Package for 6 hospitals with 400 staffed beds each and 18 non-acute satellite locations, with Dyna AI and Micromedex AI-Powered Search, the separate POISINDEX component, CareNotes Integrated, an implementation service and the RED BOOK Local Feature Advanced package updated monthly for the hospital information system. The order states 2,400 staffed beds in total. It is ready to send.

Verdict: Needs approval.

  • Large health system to review. An order for 2,400 staffed beds needs an enterprise account review before it goes out.
  • Order summary. 6 facilities with 2,400 staffed beds in total and 18 satellite locations; term 3 years.

3. Regional health system: The agent resubmits after fixing the errors

The agent's request. Please prepare a 3-year order for the DynaMedex Package for 4 hospitals with 400 staffed beds each and 18 non-acute satellite locations, with Dyna AI and Micromedex AI-Powered Search, the separate POISINDEX component, CareNotes Integrated, an implementation service and the RED BOOK Local Feature Advanced package updated monthly for the hospital information system. The order states 1,600 staffed beds in total. It is ready to send.

Verdict: Allowed.

  • Order summary. 4 facilities with 1,600 staffed beds in total and 18 satellite locations; term 3 years.

4. Teaching package: A hospital orders the Academic Preceptor package for its staff

The agent's request. Please prepare a 2-year order for the Micromedex Academic Package (Preceptor Use) for 1 hospital with 450 staffed beds. The order states 450 staffed beds in total. It is ready to send.

Verdict: Blocked.

  • Academic package for a hospital. The Academic packages are for teaching use only and cannot replace or supplement Micromedex offerings for productive use in a hospital.

5. Teaching package: The agent resubmits after fixing the errors

The agent's request. Please prepare a 2-year order for the Micromedex Clinical Evidence Plus Package for 1 hospital with 450 staffed beds. The order states 450 staffed beds in total. It is ready to send.

Verdict: Allowed.

  • Order summary. 1 facilities with 450 staffed beds in total and 0 satellite locations; term 2 years.

6. Integrated components and RED BOOK: Infobutton, embedded API and a RED BOOK J-Code feed with no integration

The agent's request. Please prepare a 3-year order for the Micromedex Clinical Evidence Plus Package for 3 hospitals with 250 staffed beds each and 6 non-acute satellite locations, with the Summary Drug Disease Lab Infobutton, the Summary Drug API for embedded use and the RED BOOK Local Feature J-Code Standard package updated monthly. The order states 750 staffed beds in total. It is ready to send.

Verdict: Blocked.

  • Integrated components need implementation. An integrated component (CareNotes Integrated, the Infobutton or the embedded API) is set up in the hospital information system; the order has no implementation service.
  • RED BOOK package and frequency not offered together. The RED BOOK package and update frequency are not offered together; Advanced is offered weekly or monthly and the J-Code packages quarterly.
  • RED BOOK needs the health information system. The RED BOOK Local Feature delivers database files for integration into the hospital health information system; the order has no such integration.

7. Integrated components and RED BOOK: The agent resubmits after fixing the errors

The agent's request. Please prepare a 3-year order for the Micromedex Clinical Evidence Plus Package for 3 hospitals with 250 staffed beds each and 6 non-acute satellite locations, with the Summary Drug Disease Lab Infobutton, the Summary Drug API for embedded use, an implementation service and the RED BOOK Local Feature J-Code Standard package updated quarterly for the hospital information system. The order states 750 staffed beds in total. It is ready to send.

Verdict: Allowed.

  • Order summary. 3 facilities with 750 staffed beds in total and 6 satellite locations; term 3 years.

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 subscription order for clinical decision support: a Micromedex or DynaMedex package for a health system, AI search add-ons, integrated components for the hospital information system and a RED BOOK data feed. Before the order goes out, the agent sends it to Predictika, which checks it against a declarative model, compares the bed count the agent states with the one it computes, 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 prepare a 3-year order for the Micromedex Clinical Evidence Premium Package for 6 hospitals. The order states 2,000 staffed beds in total. The rest of the request is on screen.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with seven findings and one approval, all in one answer. Dyna AI is an add-on that works with DynaMedex only; the order does not have the DynaMedex Package. Micromedex AI-Powered Search and Dyna AI are designed and supported for English only; the order asks for another language. The Clinical Evidence Premium Package already includes in-depth toxicology with POISINDEX; the separate POISINDEX component duplicates it. The other four are on screen.
  6. The validator returned seven blocking findings and one approval 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 order are checked against each other.
  7. What this shows: for integrated components that need an implementation service, a rules-based CPQ can do this only with workarounds.
  8. Now the agent resubmits.
  9. The corrected request. It changes the package to DynaMedex Package, the language of the AI features to english, the implementation service to yes, the RED BOOK update frequency to monthly, the used for automated claims processing to no and the stated staffed beds in total to 2,400.
  10. It calls the validator again.
  11. Not blocked, but held for a person.
  12. The seven blocking findings are cleared; one approval remains for a person: large health system to review.
  13. The agent resubmits once more.
  14. The revised request. It changes the hospitals to 4 and the stated staffed beds in total to 1,600.
  15. It calls the validator once more.
  16. This time the validator clears it.
  17. The earlier finding is cleared, and the validator's figure matches the order.
  18. Eight problems caught in one call, fixed in two passes, and a clean order. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.