Predictika · Pharmaceutical 3PL · client product onboarding

An AI agent's client product onboarding, checked before it reaches the warehouse system

An AI agent at a pharmaceutical third-party logistics provider proposes to onboard a client product at one of its distribution centers. Before it is executed, the agent sends it to Predictika, which checks it against a declarative model of the onboarding policy and returns a verdict, every reason and the figures it computed.

Scope: Larkspur Pharma Logistics (a fictional pharmaceutical 3PL) onboarding a client product at a distribution center: storage zone, free positions, DEA registration and secure storage, serialization and shelf life. See everything this validator covers

This validator models the Larkspur onboarding policy for one action, onboarding one client product at one distribution center, not just the scenarios shown below.

  • Storage — the storage zone of the product's temperature class (room temperature, refrigerated, frozen or ultra-cold) at the site, and enough free pallet positions for the forecast plus the reserve.
  • Controlled substances and serialization — the DEA registration level and the vault or cage the schedule requires, and, for prescription product, a site that is ready to handle serialized data and a client with a data feed set up.
  • Sign-off items and stated figure — utilization after onboarding and shelf life at receipt against the policy guidelines, and the number of positions the agent states against the number the validator computes.

Any client product onboarding that touches these rules is in scope, not only the scenarios pictured here. The products, clients and distribution centers are invented for illustration, the rules are simplified from public requirements and are not legal or regulatory advice, and the model does not cover pricing, receiving schedules or outbound orders.

The AI client onboarding agentBuilds the onboarding request from the conversation and says it is ready.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

6 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 client onboarding agent says

Illustrative agent statement. The validator receives the structured request below.

Proposing agent: AI client onboarding 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
The figure the agent states itselfOnly with workarounds
Positions needed is often taken as given unless custom code recomputes it from the forecast and the reserve.		The figure is recomputed from the forecast and the policy reserve and blocks on a mismatch.
Controlled substance storageOnly with care
Registration level and secure storage depend on the product schedule and on what each site actually has, so a flat product flag is not enough.		Registration level, vault and cage are each checked against the chosen product and site.
Storage zone and free spaceOnly with care
The right zone and the right free count depend on the product temperature class, and free space has to include the reserve.		The class totals are picked from the product, then compared with the forecast plus the reserve.
Serialization for prescription productOnly with care
The rule applies only to prescription product, and depends on both the site and the client data feed.		Site readiness and the client data feed are checked together, and only for prescription product.
Utilization limits and shelf lifeOnly with workarounds
A threshold that applies only when space is available, and a separate one for shelf life, are easy to get wrong at the boundary.		Utilization is computed after onboarding and compared with the limit; each boundary was run on the live validator.
Every problem returned in one answerOnly with workarounds
Rules tend to stop at the first failing check, so reasons surface one resubmission at a time.		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 request; approval findings need a person to sign off; everything else is computed.

    
1DEA registration level too low: Extended-Release Pain Tablets 60ct needs registration for the schedule level 2 and Allentown DC is registered for level 1.
2Controlled substance storage: Extended-Release Pain Tablets 60ct needs secure storage for schedule level 2 and Allentown DC does not have it.
3Stated positions do not match: the agent stated 250 positions; the validator computes 275.
4Site would be very full: 90.7 percent of this storage class would be in use; sign-off is needed above 85 percent.
5Not enough free space: 66 positions are needed including the reserve; Dallas DC has 40 free.
6Site not serialization-ready: Insulin Pens 5-pack is a prescription product and Dallas DC cannot yet handle serialized product data.
7No serialization data feed: Insulin Pens 5-pack is a prescription product and Quillon Generics has no transaction data feed set up.
8Stated positions do not match: the agent stated 60 positions; the validator computes 66.
9Short shelf life at receipt: 55 percent remains; sign-off is needed below 60 percent.
10DEA registration level too low: Anxiety Tablets 30ct needs registration for the schedule level 1 and Reno DC is registered for level 0.
11Controlled substance storage: Anxiety Tablets 30ct needs secure storage for schedule level 1 and Reno DC does not have it.
12Site would be very full: 100 percent of this storage class would be in use; sign-off is needed above 85 percent.
13Not enough free space: 44 positions are needed including the reserve; Allentown DC has 33 free.
14No storage zone: Gene Therapy Vials is a ultracold product and Louisville DC has no ultracold space.
15No storage zone: Frozen Vaccine Vials is a frozen product and Reno DC has no frozen space.
16Site would be very full: 85.3 percent of this storage class would be in use; sign-off is needed above 85 percent.
17Short shelf life at receipt: 59 percent remains; sign-off is needed below 60 percent.
18Stated positions do not match: the agent stated 100 positions; the validator computes 110.
19Site not serialization-ready: Cardiovascular Tablets 30ct is a prescription product and Dallas DC cannot yet handle serialized product data.
20No serialization data feed: Cardiovascular Tablets 30ct is a prescription product and Aldervale Therapeutics has no transaction data feed set up.

  

  

    
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. Prescription insulin pens: Onboarding of insulin pens at a distribution center that is nearly full and not ready for serialized data, for a client with no data feed

    
The agent's request. Please onboard Insulin Pens 5-pack for Quillon Generics at Dallas DC, with a forecast of 60 pallets, so 60 storage positions are needed. The product will arrive with 80 percent of its shelf life. It is ready to execute.

    
Verdict: Blocked.

    
    
      
  • Not enough free space. Not enough free space: 66 positions are needed including the reserve; Dallas DC has 40 free.
    • 
      
  • Site not serialization-ready. Site not serialization-ready: Insulin Pens 5-pack is a prescription product and Dallas DC cannot yet handle serialized product data.
    • 
      
  • No serialization data feed. No serialization data feed: Insulin Pens 5-pack is a prescription product and Quillon Generics has no transaction data feed set up.
    • 
      
  • Stated positions do not match. Stated positions do not match: the agent stated 60 positions; the validator computes 66.
    • 
    

    
2. Prescription insulin pens: The agent resubmits after fixing the errors

    
The agent's request. Please onboard Insulin Pens 5-pack for Quillon Generics at Allentown DC, with a forecast of 60 pallets, so 66 storage positions are needed. The product will arrive with 80 percent of its shelf life. It is ready to execute.

    
Verdict: Blocked.

    
    
      
  • No serialization data feed. No serialization data feed: Insulin Pens 5-pack is a prescription product and Quillon Generics has no transaction data feed set up.
    • 
    

    
3. Prescription insulin pens: The agent resubmits after fixing the errors

    
The agent's request. Please onboard Insulin Pens 5-pack for Quillon Generics at Allentown DC, with a forecast of 60 pallets, so 66 storage positions are needed. The product will arrive with 80 percent of its shelf life. The client has its serialization data feed set up. It is ready to execute.

    
Verdict: Allowed.

    
    
      
  • Computed figures. Positions needed 66; utilization after onboarding 77 percent.
    • 
    

    
4. Controlled substance: Onboarding of Schedule II pain tablets at a distribution center registered for a lower DEA level

    
The agent's request. Please onboard Extended-Release Pain Tablets 60ct for Aldervale Therapeutics at Allentown DC, with a forecast of 250 pallets, so 250 storage positions are needed. The product will arrive with 80 percent of its shelf life. The client has its serialization data feed set up. It is ready to execute.

    
Verdict: Blocked.

    
    
      
  • DEA registration level too low. DEA registration level too low: Extended-Release Pain Tablets 60ct needs registration for the schedule level 2 and Allentown DC is registered for level 1.
    • 
      
  • Controlled substance storage. Controlled substance storage: Extended-Release Pain Tablets 60ct needs secure storage for schedule level 2 and Allentown DC does not have it.
    • 
      
  • Stated positions do not match. Stated positions do not match: the agent stated 250 positions; the validator computes 275.
    • 
      
  • Site would be very full. Site would be very full: 90.7 percent of this storage class would be in use; sign-off is needed above 85 percent.
    • 
    

    
5. Controlled substance: The agent resubmits after fixing the errors

    
The agent's request. Please onboard Extended-Release Pain Tablets 60ct for Aldervale Therapeutics at Louisville DC, with a forecast of 250 pallets, so 275 storage positions are needed. The product will arrive with 80 percent of its shelf life. The client has its serialization data feed set up. It is ready to execute.

    
Verdict: Allowed.

    
    
      
  • Computed figures. Positions needed 275; utilization after onboarding 84.4 percent.
    • 
    

    
6. Shelf life at receipt: Onboarding of allergy tablets arriving with a short shelf life

    
The agent's request. Please onboard Allergy Relief Tablets 90ct for Halden Consumer Health at Dallas DC, with a forecast of 600 pallets, so 660 storage positions are needed. The product will arrive with 55 percent of its shelf life. It is ready to execute.

    
Verdict: Needs approval.

    
    
      
  • Short shelf life at receipt. Short shelf life at receipt: 55 percent remains; sign-off is needed below 60 percent.
    • 
      
  • Computed figures. Positions needed 660; utilization after onboarding 63.2 percent.
    • 
    

  

  

    
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 agent at a pharmaceutical third-party logistics provider proposes to onboard a client product at one of its distribution centers. Before it is executed, the agent sends it to Predictika, which checks it against a declarative model of the onboarding policy and returns a verdict, every reason and the figures it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please onboard Insulin Pens 5-pack for Quillon Generics at Dallas DC, with a forecast of 60 pallets, so 60 storage positions are needed. The product will arrive with 80 percent of its shelf life. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with four findings, all in one answer. Not enough free space: 66 positions are needed including the reserve; Dallas DC has 40 free. Site not serialization-ready: Insulin Pens 5-pack is a prescription product and Dallas DC cannot yet handle serialized product data. No serialization data feed: Insulin Pens 5-pack is a prescription product and Quillon Generics has no transaction data feed set up. The other one is on screen.
    6. 
      
  6. The validator returned four 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 request are checked against each other.
    7. 
      
  7. What this shows: for storage zone and free space, a rules-based CPQ is error-prone for this.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the distribution center to Allentown DC and the stated storage positions to 66.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Still blocked, with one finding, all in one answer. No serialization data feed: Insulin Pens 5-pack is a prescription product and Quillon Generics has no transaction data feed set up.
    12. 
      
  12. Three findings cleared; still blocked on no serialization data feed.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It changes the client serialization data feed to set up.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. The earlier finding is cleared, and the validator's figure matches the request.
    18. 
      
  18. Four problems caught in one call, fixed in two passes, and a clean request. That is the value of Predictika's validator: catching mistakes before the action runs.
    19.