Predictika · Emulsifier blend quotes

An AI sales agent's emulsifier blend quote, checked before it goes out

An AI sales agent prepares an emulsifier blend quote for a customer's batch: a Croda Span and a Tween, the share of each, the oil to emulsify, the emulsifier level and the weight needed. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how the blend has to match the oil and returns a verdict, every reason and the blend HLB and emulsifier weight it computed.

Scope: Croda Span and Tween emulsifier blends for oil-in-water and water-in-oil emulsions: seven Span and seven Tween grades, five oils, the blend HLB, the emulsifier level and the weight for the batch. See everything this validator covers

This validator models Croda Span and Tween blend quotes, not just the scenarios shown below.

  • Grades and blend — Span 20, 40, 60, 80, 83, 85 and 120 with Tween 20, 21, 40, 60, 61, 65 and 80, and the share of each in the blend; the blend HLB is computed from the published HLB value of each grade.
  • Oil and emulsion type — paraffinic mineral oil, rapeseed oil, isopropyl myristate, castor oil and cetyl alcohol, each with the HLB it needs, for oil-in-water or water-in-oil emulsions, each with its own HLB window.
  • Level and weight — the emulsifier level as a percent of the oil phase, and the emulsifier weight for the batch, compared with the figures the quote states.
  • Approvals — a blend HLB near the edge of the tolerance, a Span and a Tween from different fatty-acid families, and a high emulsifier level go to a formulator.

Any Croda Span and Tween blend quote that touches these rules is in scope — not only the scenarios pictured here. Not covered: other Croda product lines, pack sizes and prices; no prices are modeled. The product names, HLB values and the Span 80 with Tween 80 blend for paraffinic mineral oil come from the public Croda datasheet; the HLB the other oils need, the tolerances, the HLB windows and the level limits are modeled from comparable products and are not published by Croda.

The AI sales agentBuilds the emulsifier blend quote from the conversation with the customer 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 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
A blend HLB computed from two grades and their sharesOnly with care
The blend HLB depends on both published values and the share of each, so a rule has to recompute it for every pair and compare it with the HLB the oil needs.		The two grades, their shares and the oil are in one request and the blend HLB is computed and compared on every call.
A tolerance that depends on the emulsion typeOnly with care
An oil-in-water and a water-in-oil emulsion each need their own HLB window, so every window needs its own rule that reads the emulsion type.		The emulsion type is a field of the request and selects the window on every call.
Grades that should come from the same fatty-acid familyOnly with workarounds
A mismatch is allowed with a sign-off, so a family lookup and an approval rule are needed beside the block rules.		The family of each grade is looked up in the model and a mismatch goes to a formulator on every call.
Stated figures that must match the computed onesOnly with workarounds
The blend HLB and the emulsifier weight are built from several fields, so a separate rule is needed to recompute each and compare it with the stated figure.		The stated blend HLB and weight are fields of the request and are compared with the computed figures 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 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 quote; approval findings need a person to sign off; everything else is computed.

    
1The blend HLB of 10.7 is more than 0.5 from the oil requirement of 10.0; a formulator should confirm.
2The blend HLB is 11.8; the oil needs 10.0, and this model allows 1.0 of deviation.
3A Span and Tween blend needs both parts; the Span share of 0 percent leaves one of them out.
4The blend HLB is 15.0; the oil needs 10.0, and this model allows 1.0 of deviation.
5A Span and Tween blend needs both parts; the Span share of 100 percent leaves one of them out.
6The blend HLB is 4.3; the oil needs 10.0, and this model allows 1.0 of deviation.
7An oil-in-water emulsion needs a blend HLB of at least 8.0; this blend is 4.3.
8A water-in-oil emulsion needs a blend HLB between 3.0 and 8.0; this blend is 10.0.
9The emulsifier level is 4 percent of the oil phase; this model needs at least 5 percent (about 10 percent is a typical starting point).
10The emulsifier level of 25 percent of the oil phase is high; a formulator should confirm.
11The quote states a blend HLB of 12.0; the chosen Span and Tween shares give 10.0.
12The quote states 25 kg of emulsifier; the batch needs 20 kg.
13The blend HLB is 13.0; the oil needs 14.0, and this model allows 1.0 of deviation.
14The Span and Tween chosen are from different fatty-acid families; matching pairs (for example Span 80 with Tween 80, Span 60 with Tween 60) are the usual practice, so a formulator should confirm.
15The blend HLB is 13.7; the oil needs 15.5, and this model allows 1.0 of deviation.

  

  

    
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. Blend off target and level too low: A 2,000 kg paraffinic mineral oil emulsion blended from Span 80 and Tween 80, with a low emulsifier level

    
The agent's request. Please quote an oil-in-water emulsion of paraffinic mineral oil for a 2,000 kg batch, 25 percent oil, blended from Span 80 at 30 percent and Tween 80 at 70 percent, with the emulsifier at 4 percent of the oil phase, stating 20 kg of emulsifier and a blend HLB of 11.6. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Blend HLB too far from the oil requirement. The blend HLB is 11.8; the oil needs 10.0, and this model allows 1.0 of deviation.
    • 
      
  • Emulsifier level too low. The emulsifier level is 4 percent of the oil phase; this model needs at least 5 percent (about 10 percent is a typical starting point).
    • 
      
  • Stated blend HLB does not match. The quote states a blend HLB of 11.6; the chosen Span and Tween shares give 11.8.
    • 
    

    
2. Blend off target and level too low: The agent resubmits after fixing the errors

    
The agent's request. Please quote an oil-in-water emulsion of paraffinic mineral oil for a 2,000 kg batch, 25 percent oil, blended from Span 80 at 47 percent and Tween 80 at 53 percent, with the emulsifier at 10 percent of the oil phase, stating 50 kg of emulsifier and a blend HLB of 10.0. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Blend summary. Blend HLB 10.0 (oil needs 10.0), 50 kg of emulsifier for a 2,000 kg batch.
    • 
    

    
3. Mixed fatty-acid families: A 1,000 kg rapeseed oil emulsion blended from Span 60 and Tween 20

    
The agent's request. Please quote an oil-in-water emulsion of rapeseed oil for a 1,000 kg batch, 20 percent oil, blended from Span 60 at 60 percent and Tween 20 at 40 percent, with the emulsifier at 10 percent of the oil phase, stating 20 kg of emulsifier and a blend HLB of 9.5. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Blend HLB too far from the oil requirement. The blend HLB is 9.5; the oil needs 8.0, and this model allows 1.0 of deviation.
    • 
      
  • Span and Tween from different fatty-acid families. The Span and Tween chosen are from different fatty-acid families; matching pairs (for example Span 80 with Tween 80, Span 60 with Tween 60) are the usual practice, so a formulator should confirm.
    • 
    

    
4. Mixed fatty-acid families: The agent resubmits after fixing the errors

    
The agent's request. Please quote an oil-in-water emulsion of rapeseed oil for a 1,000 kg batch, 20 percent oil, blended from Span 60 at 66 percent and Tween 60 at 34 percent, with the emulsifier at 10 percent of the oil phase, stating 20 kg of emulsifier and a blend HLB of 8.2. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Blend summary. Blend HLB 8.2 (oil needs 8.0), 20 kg of emulsifier for a 1,000 kg batch.
    • 
    

    
5. Water-in-oil emulsion: A 1,000 kg water-in-oil emulsion of rapeseed oil blended from Span 60 and Tween 20

    
The agent's request. Please quote a water-in-oil emulsion of rapeseed oil for a 1,000 kg batch, 20 percent oil, blended from Span 60 at 20 percent and Tween 20 at 80 percent, with the emulsifier at 10 percent of the oil phase, stating 20 kg of emulsifier and a blend HLB of 14.3. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Water-in-oil emulsion needs a low HLB. A water-in-oil emulsion needs a blend HLB between 3.0 and 8.0; this blend is 14.3.
    • 
      
  • Span and Tween from different fatty-acid families. The Span and Tween chosen are from different fatty-acid families; matching pairs (for example Span 80 with Tween 80, Span 60 with Tween 60) are the usual practice, so a formulator should confirm.
    • 
    

    
6. Water-in-oil emulsion: The agent resubmits after fixing the errors

    
The agent's request. Please quote a water-in-oil emulsion of rapeseed oil for a 1,000 kg batch, 20 percent oil, blended from Span 60 at 75 percent and Tween 60 at 25 percent, with the emulsifier at 10 percent of the oil phase, stating 20 kg of emulsifier and a blend HLB of 7.3. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Blend summary. Blend HLB 7.3 (oil needs 8.0), 20 kg of emulsifier for a 1,000 kg batch.
    • 
    

  

  

    
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 an emulsifier blend quote for a customer's batch: a Croda Span and a Tween, the share of each, the oil to emulsify, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how the blend has to match the oil and returns a verdict, every reason and the blend HLB and emulsifier weight it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote an oil-in-water emulsion of paraffinic mineral oil for a 2,000 kg batch, 25 percent oil, blended from Span 80 at 30 percent and Tween 80 at 70 percent. It is ready to send. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with three findings, all in one answer. The blend HLB is 11.8; the oil needs 10.0, and this model allows 1.0 of deviation. The emulsifier level is 4 percent of the oil phase; this model needs at least 5 percent (about 10 percent is a typical starting point). The quote states a blend HLB of 11.6; the chosen Span and Tween shares give 11.8.
    6. 
      
  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.
    7. 
      
  7. What this shows: for a blend HLB computed from two grades and their shares, a rules-based CPQ is error-prone for this.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the span share to 47 percent, the tween share to 53 percent, the emulsifier level to 10 percent of the oil phase, the Blend HLB stated on the quote to 10 and the emulsifier stated on the quote to 50 kg.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. This time the validator clears it.
    12. 
      
  12. All three earlier findings are cleared.
    13. 
      
  13. Three problems caught in one call, fixed in one pass, 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.
    14.