Predictika · Lithography scanner orders

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

An AI sales agent prepares a quote for TWINSCAN DUV and EUV lithography scanners, in euros. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the configuration and commercial rules, compares the total the quote states with the computed one, and returns every finding together with the line-item quote. This is an illustrative model built from public product information with estimated prices; it is not ASML's internal rule set.

Scope: orders for TWINSCAN DUV and EUV lithography scanners (NXT:870, 1470, 2000i, 2100i, NXE:3800E and EXE:5200), with options, installation and service. See everything this validator covers

This validator models one slice of semiconductor equipment sales: an order for lithography scanners with its options, installation and qualification, and a service agreement. It handles any combination of these scanner models, nodes, regions and service terms, not only the scenarios shown on this page. It does not cover metrology or other product lines, real prices, financing, freight or tax.

Covered: fit to the process — resolution against the feature the layer needs, multi-patterning, effective throughput, scanner count from the wafer plan, overlay and the productivity package.
Covered: the fab and the rules around it — facility power, export control by region, EUV service, service term, lead time and production slot swaps.
Covered: commercial terms — discount authority by order size, the line-item quote, and the total the agent states.
Labeled as assumptions — every price, specification and limit is an illustrative estimate based on public information, not an ASML list.

The AI sales agentBuilds the scanner order quote 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 sales agent says

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 caught	Typical rules-based CPQ	Predictika
Scanner count from the wafer plan	Only with workarounds The count depends on wafer starts, layers and an effective throughput that changes with multi-patterning and the productivity package, so a rules-based tool needs a table for each combination.	The count is computed once from the plan and the effective throughput, and the finding states what is needed against what is quoted.
Node reach with multi-patterning	Only with care Whether a scanner can print a feature depends on its type and on the multi-patterning choice, and the same choice halves the throughput.	Resolution, throughput and scanner count are derived from the same choice, so the effects stay consistent.
Export control by region and class	Only with workarounds The outcome is a block for some pairs and an approval for others, which a rules-based tool handles as one long list of pair rules.	Region and export class are attributes of the quote and the scanner, and the validator returns a block or an approval with the reason.
Delivery date against lead time	Only with care A date can be fine, need a production slot swap, or be impossible, and each outcome routes differently.	The three outcomes come from one comparison, and the approval is listed as a finding rather than a failed check.
Discount authority by order size	Only with care The allowed discount and the approver both depend on the number of scanners.	The tier is derived from the quantity, and a discount between the account manager limit and the tier maximum is an approval.
Total the agent states	Only with workarounds A validation rule can compare a stated figure with a calculated one only if the stated figure is a field on the quote, which it rarely is.	The stated total travels with the quote and is compared with the computed total on every call; a mismatch is its own block.

What this validator catches: findings from its test cases