Predictika · Grocery and retail checkout

An AI shopping agent's checkout basket, checked before it goes out

An AI shopping agent assembles a checkout basket for a drugstore or grocery shopper: items, manufacturer and store coupons, an order coupon and ExtraBucks. Before the basket goes out, the agent sends it to Predictika, which checks every coupon against a declarative model of the coupon policy, compares the net total the agent states with the one it computes, and returns every finding together with the line-item view.

Scope: drugstore and grocery checkout baskets of up to 3 lines: items, manufacturer and store coupons, an order coupon and ExtraBucks. See everything this validator covers

This is a validator for one thing: whether the coupons in a checkout basket stack legally under a typical drugstore coupon policy, and whether the net total the agent states is right. It does not cover loyalty programs, tax, or any retailer's actual policy text.

  • Per item — one manufacturer and one store coupon per item, the right kind in the right slot, brand, category and size match, expiration, buy-N quantities, and the coupon value capped at the item price.
  • Across the basket — no more than 4 identical manufacturer coupons in a transaction, and order coupons ($5 off $25, $10 off $50) counted after item coupons without alcohol, gift cards or prescriptions.
  • Rewards and totals — ExtraBucks limited to the eligible items left after coupons, a line-item view of the whole basket, and the net total the agent states.

Any basket built from these items, coupons and order coupons runs through this validator, not just the scenarios shown on this page. It is an illustrative model; items, prices, coupons and limits are fictional.

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

8 scenarios in 4 groups · each group runs from the agent’s first attempt to the outcome · click one to switch

What the agent proposes

The AI shopping agent says

Proposing agent: AI shopping 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 typical CPQ products) 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
One manufacturer and one store coupon per item, with the coupon kind checked against the slot it was placed inOnly with care
Each slot needs its own rule against the coupon type; a coupon scanned into the wrong slot has to be caught by a rule written for exactly that case.		The coupon record carries its kind, and every slot is checked against it on every call, together with every other rule.
An order coupon threshold counted after item coupons, with alcohol, gift cards and prescriptions left outOnly with workarounds
The threshold sees the cart total, so the items that do not count and the item coupons have to be subtracted by hand in a chain of calculated fields.		The qualifying total is derived on each call from the line items, their categories and the item coupons, then compared with the threshold.
Coupon value capped at the item price, with the cap spread over a manufacturer and a store couponOnly with care
The cap and the approval for a coupon worth more than the item need their own fields on every line.		Each line derives its raw coupon value, its cap and its overage, and the validator lists the line for approval.
Identical manufacturer coupons limited across the whole transaction, and buy-two coupons that need two items on the same lineOnly with care
The limit spans lines, so a rule per line misses the total; each buy-N coupon needs its own quantity rule.		Coupon counts are summed across the whole basket and compared with the item quantities in one pass.
A stated net total that does not match the computed totalOnly with care
A validation rule can compare a stated total with a calculated field, but only if the stated total is a field on the basket, which it rarely is.		The stated net total travels with the basket and is compared with the computed one 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 basket; approval findings need a person to sign off; everything else is computed.

    
1Two manufacturer coupons on one item (policy allows one manufacturer and one store coupon per item).
2Manufacturer coupon in the store slot, or the other way round.
3Coupon used after its expiration date.
4Coupon brand, category or size does not match the item.
5Buy-N coupon without enough items on the line.
6More than 4 identical manufacturer coupons in one transaction.
7Order coupon threshold not met after item coupons (alcohol, gift cards and prescriptions do not count).
8Order coupon expired.
9ExtraBucks above the eligible items left after coupons.
10Stated net total differs from the computed total by more than $1.00.
11Coupons worth more than the item price (approval; the discount is cut to the item price, no cash back).