Predictika · Colony health monitoring submissions

An AI sales agent's colony health monitoring submission, checked before it goes out

An AI sales agent prepares a health monitoring submission for a research animal colony: which agents to test for, by which method, on which samples, and how many samples are pooled into one test. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of three published data sheets and returns a verdict, every reason and the figures it computed.

Scope: Full Spectrum Health Monitoring submissions for mouse, rat, hamster, guinea pig and rabbit colonies: 91 agents across bacteriology, parasitology and virus testing. See everything this validator covers

This validator models submissions for the Full Spectrum Health Monitoring service as described in three public data sheets (bacteriology, parasitology and virus testing), not just the scenarios shown below.

  • Agents and species — each of the 91 agents is listed on the data sheets for particular species; the validator checks the agent against the species of the colony.
  • Test methods and sample types — each agent has its own list of test methods and of sample types (live animal, serum, dried blood spot, fecal, environmental, oral swab, fur swab); the validator checks both. The sheets say sample types can be species specific without saying which, so only the agent's listed types are checked.
  • Pooling — environmental samples, and fecal samples for rtPCR, may be pooled up to 10 into one test (blocked above); parasitology fur swabs are recommended at 5 or fewer and accepted up to 10; bacteriology oral and fur swabs are recommended at 2 or fewer; pooling serum or dried blood spots for virus tests is not recommended. Recommendations go to a person for approval.
  • Counts — tests times samples per test, added over up to three test lines, is compared with the sample count the submission states.

Any health monitoring submission that touches these rules is in scope — not only the scenarios pictured here. Not covered: prices (the data sheets give none, so there is no price build-up or line-item list), the agents Entamoeba spp and Simian Virus (their rows carry no species marks on the data sheets), which test method suits which sample type, reporting times, sampling kits, pathology, animal transportation, testing of biological materials, the colony monitoring programs for animals Inotiv sells, and any agent not on the three data sheets.

The AI sales agentBuilds the health monitoring submission 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
An agent that is not listed for the colony's speciesOnly with care
The sheets are a matrix of 91 agents against five species; a CPQ can hold it as compatibility rules, but every sheet revision means editing the matrix.		The species and the agent are in one request and are compared against the matrix on every call.
A test method the agent is not offered withOnly with care
Each agent has its own list of methods; a CPQ handles it with option lists per agent, maintained by hand.		Agent and method arrive together and are checked on every call.
A sample type the agent is not offered withOnly with care
Another per-agent list, with a different set of sample types for each agent.		Agent and sample type are checked together.
Pooling above a limit that depends on sample type and methodOnly with workarounds
Environmental samples are capped at 10 for every method, fecal samples only for rtPCR and parasite fur swabs by their own limit, so each limit needs a combination rule.		Sample type, method, agent family and pool size are checked together.
A recommended pooling level that needs a person's approval, not a blockOnly with workarounds
A rules-based CPQ treats a rule as pass or fail; a soft limit needs approval routing built next to the hard limits.		Hard limits block and recommended limits go to approval, in the same call.
A stated sample count that must match the linesOnly 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 it.		The stated count travels with the submission and is compared with the computed count on every call.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as a step is completed, so a rep 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.

    
1Test line 1 asks for Rat Theilo Virus on a mouse colony; the Inotiv data sheet does not list that agent for that species.
2Test line 1 asks for Mouse Hepatitis Virus by Culture; the Inotiv data sheet does not list that test method for that agent.
3Test line 1 submits serum samples for Mouse Kidney Parvovirus; the Inotiv data sheet does not list that sample type for that agent.
4Test line 2 has 0 tests of 5 samples each; both numbers must be at least 1.
5Test line 3 has 6 tests of 0 samples each; both numbers must be at least 1.
6Test line 1 pools 11 environmental samples into one test; the Inotiv data sheets allow at most 10.
7Test line 2 pools 11 fecal samples into one rtPCR test; the Inotiv data sheets allow at most 10.
8Test line 1 pools 6 fur swabs into one parasitology test; the Inotiv data sheet recommends no more than 5 but accepts pools up to 10, so a person should confirm.
9Test line 1 pools 11 fur swabs into one parasitology test; the Inotiv data sheet accepts pools up to 10.
10Test line 1 pools 3 oral or fur swabs into one bacteriology test; the Inotiv data sheet recommends up to 2, so a person should confirm.
11Test line 1 pools 2 serum or dried blood spot samples into one virus test; the Inotiv data sheet does not recommend pooling these, so a person should confirm.
12The submission has no test lines; at least one agent must be selected.
13The submission states 71 samples; the test lines add up to 72 samples.
14Test line 3 asks for Syphacia muris by ELISA; the Inotiv data sheet does not list that test method for that agent.
15Test line 3 submits fur swab samples for Giardia spp; the Inotiv data sheet does not list that sample type for that agent.
16Test line 3 pools 7 fur swabs into one parasitology test; the Inotiv data sheet recommends no more than 5 but accepts pools up to 10, so a person should confirm.
17Test line 2 pools 4 oral or fur swabs into one bacteriology test; the Inotiv data sheet recommends up to 2, so a person should confirm.

  

  

    
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. Mouse virus panel: A mouse colony virus panel

    
The agent's request. Please run a health monitoring submission for our mouse colony: line 1: 12 tests of Mouse Hepatitis Virus by MFIA on serum samples, 1 sample per test; line 2: 6 tests of Mouse Kidney Parvovirus by ELISA on serum samples, 1 sample per test; line 3: 6 tests of Rat Theilo Virus by ELISA on serum samples, 1 sample per test. The submission covers 24 samples in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Sample type not listed for the agent (line 2). Test line 2 submits serum samples for Mouse Kidney Parvovirus; the Inotiv data sheet does not list that sample type for that agent.
    • 
      
  • Agent not listed for the species (line 3). Test line 3 asks for Rat Theilo Virus on a mouse colony; the Inotiv data sheet does not list that agent for that species.
    • 
    

    
2. Mouse virus panel: The agent resubmits after fixing the errors

    
The agent's request. Please run a health monitoring submission for our mouse colony: line 1: 12 tests of Mouse Hepatitis Virus by MFIA on serum samples, 1 sample per test; line 2: 6 tests of Mouse Kidney Parvovirus by ELISA on fecal samples, 1 sample per test; line 3: 6 tests of Mouse Noro Virus by ELISA on serum samples, 1 sample per test. The submission covers 24 samples in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Submission summary. 3 test line(s) and 24 sample(s) in total.
    • 
    

    
3. Rat colony pooling: A rat colony submission with pooled samples

    
The agent's request. Please run a health monitoring submission for our rat colony: line 1: 8 tests of Rodentibacter pneumotropicus by rtPCR on environmental samples, 12 samples per test; line 2: 10 tests of Syphacia muris by Microscope on fecal samples, 10 samples per test; line 3: 6 tests of Staphylococcus aureus by Culture on oral swab samples, 4 samples per test. The submission covers 220 samples in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Environmental pool above 10 (line 1). Test line 1 pools 12 environmental samples into one test; the Inotiv data sheets allow at most 10.
    • 
      
  • Swab pool above the recommended 2 (line 3). Test line 3 pools 4 oral or fur swabs into one bacteriology test; the Inotiv data sheet recommends up to 2, so a person should confirm.
    • 
    

    
4. Rat colony pooling: The agent resubmits after fixing the errors

    
The agent's request. Please run a health monitoring submission for our rat colony: line 1: 8 tests of Rodentibacter pneumotropicus by rtPCR on environmental samples, 10 samples per test; line 2: 10 tests of Syphacia muris by Microscope on fecal samples, 10 samples per test; line 3: 6 tests of Staphylococcus aureus by Culture on oral swab samples, 4 samples per test. The submission covers 204 samples in total. It is ready to send.

    
Verdict: Needs approval.

    
    
      
  • Swab pool above the recommended 2 (line 3). Test line 3 pools 4 oral or fur swabs into one bacteriology test; the Inotiv data sheet recommends up to 2, so a person should confirm.
    • 
      
  • Submission summary. 3 test line(s) and 204 sample(s) in total.
    • 
    

    
5. Rat colony pooling: The agent resubmits after fixing the errors

    
The agent's request. Please run a health monitoring submission for our rat colony: line 1: 8 tests of Rodentibacter pneumotropicus by rtPCR on environmental samples, 10 samples per test; line 2: 10 tests of Syphacia muris by Microscope on fecal samples, 10 samples per test; line 3: 6 tests of Staphylococcus aureus by Culture on oral swab samples, 2 samples per test. The submission covers 192 samples in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Submission summary. 3 test line(s) and 192 sample(s) in total.
    • 
    

    
6. Rabbit colony: A rabbit colony virus panel with Salmonella

    
The agent's request. Please run a health monitoring submission for our rabbit colony: line 1: 10 tests of Rabbit Haemorrhagic Disease Virus by ELISA on serum samples, 1 sample per test; line 2: 10 tests of Sendai Virus by ELISA on serum samples, 1 sample per test; line 3: 8 tests of Salmonella spp. by Culture on fecal samples, 1 sample per test. The submission covers 28 samples in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Agent not listed for the species (line 2). Test line 2 asks for Sendai Virus on a rabbit colony; the Inotiv data sheet does not list that agent for that species.
    • 
    

    
7. Rabbit colony: The agent resubmits after fixing the errors

    
The agent's request. Please run a health monitoring submission for our rabbit colony: line 1: 10 tests of Rabbit Haemorrhagic Disease Virus by ELISA on serum samples, 1 sample per test; line 2: 10 tests of Rabbit Rotavirus by ELISA on serum samples, 1 sample per test; line 3: 8 tests of Salmonella spp. by Culture on fecal samples, 1 sample per test. The submission covers 28 samples in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Submission summary. 3 test line(s) and 28 sample(s) in total.
    • 
    

  

  

    
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 health monitoring submission for a research animal colony: which agents to test for, by which method, on which samples, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of three published data sheets 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 run a health monitoring submission for our mouse colony: line 1: 12 tests of Mouse Hepatitis Virus by MFIA on serum samples. The submission covers 24 samples in total. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with two findings, all in one answer. Test line 2 submits serum samples for Mouse Kidney Parvovirus; the Inotiv data sheet does not list that sample type for that agent. Test line 3 asks for Rat Theilo Virus on a mouse colony; the Inotiv data sheet does not list that agent for that species.
    6. 
      
  6. The validator returned two 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 order are checked against each other.
    7. 
      
  7. What this shows: for an agent that is not listed for the colony's species, a rules-based CPQ is error-prone for this.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the test line 2 to 6 tests of Mouse Kidney Parvovirus by ELISA on fecal samples, 1 sample per test and the test line 3 to 6 tests of Mouse Noro Virus by ELISA on serum samples, 1 sample per test.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. Both earlier findings are cleared, and the validator's figure matches the order.
    13. 
      
  13. Two problems caught in one call, fixed in one pass, 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.
    14.