Predictika · KLA inspection, review and metrology quotes

An AI sales agent's KLA tool quote, checked before it goes out

An AI sales agent prepares a quote for wafer inspection, e-beam review or overlay metrology tools, with sensitivity package, handler, Klarity and service. Before it goes out, the agent sends it to Predictika, which checks the tool against the customer's process node, throughput requirement, destination and delivery date, compares the total and the throughput the quote states with the computed ones, and returns every finding together with the line-item quote. This is an illustrative model built from public product information with invented prices and throughputs; it is not KLA's internal rule set.

Scope: KLA wafer inspection, e-beam review and overlay metrology tools: the 29xx and 39xx broadband inspectors, Puma, Surfscan SP7, eDR7xxx e-beam review and Archer overlay metrology, with sensitivity package, handler, Klarity and service. See everything this validator covers

This validator models one tool set: the KLA 29xx and 39xx broadband patterned inspectors, Puma, Surfscan SP7, eDR7xxx e-beam review and Archer overlay metrology, quoted with the high-sensitivity package, a 2-port or 4-port handler, Klarity defect data management and a service plan. It handles any quote involving these tools, not only the scenarios shown on this page.

Covered: tool fit — each tool's smallest supported process node, the high-sensitivity package (inspectors only), which cuts throughput and adds to the price, and the throughput of the whole fleet against what the customer requires.
Covered: handling and prerequisites — the 4-load-port handler for tools above a throughput threshold, Klarity defect data management for fleets of 3 or more tools, and an inspector on site for e-beam review.
Covered: export, service and delivery — export-controlled tools blocked, and any restricted destination sent for trade-compliance review; Premium service needing a term of at least 2 years and service terms of 1 to 5 years; tool lead time against the customer's need-by date; and discount authority, which is higher for fleets of 3 or more tools.
Covered: the quote itself — the line-item quote, and the total and fleet throughput the agent states, which must match what the validator computes.
Not covered — other KLA product lines and tools outside this set, software sold on its own, spares and training, which specific inspector model must accompany a review tool (the model only requires that an inspector is on site), financing and tax. Other areas can be added.
Labeled as assumptions — every price, throughput, minimum process node, lead time, export-control flag, service rate, handler and Klarity price, and discount limit is an invented illustrative value, not KLA data.

The AI sales agentBuilds the KLA tool 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.

7 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.

What this means

Run the validator to see the outcome

The explanation appears here once the validator returns.

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
A package that changes throughput, which then drives the handler and the fleet check	Only with workarounds The high-sensitivity package cuts wafers per hour, and that figure decides both the handler and whether the fleet meets the requirement. In a CPQ it takes custom throughput fields and a chain of rules, and a missed step lets a fleet that is too slow through.	Throughput after the package is computed once; the handler threshold and the fleet check both read it.
Klarity and the discount ceiling that depend on fleet size	Only with care Validation rules that read the total quantity are possible, but each prerequisite and each ceiling is its own rule that someone has to write and keep in step.	Fleet size is part of the model, so Klarity and the discount ceiling change together.
E-beam review that needs an inspector at the customer site	Only with workarounds The inspector installed at the site is not on the quote, so the rule needs data from outside the order, usually a lookup that is built and maintained separately.	Whether an inspector is on site travels with the quote and is checked on every call.
Export-controlled tool to a restricted destination, versus a tool that only needs review	Only with care Possible with a pair of rules per destination, but it is easy to block everything or to wave through a tool that should have been stopped.	The tool and the destination are compared; one is a block, the other an approval, and both are reported.
A stated total or throughput that does not match the computed figure	Only with care A validation rule can compare a stated figure with a calculated field, but only if the stated figure is a field on the quote, which it rarely is.	Both stated figures travel with the quote and are compared with the computed ones 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 quote; approval findings need a person to sign off; everything else is computed.

1	3 tools at 60 wafers per hour each deliver 180 wafers per hour, short of the 200 required.
2	90 wafers per hour per tool needs the 4-load-port handler.
3	A discount of 12 percent is above the 5 percent an agent may give alone.
4	A discount of 16 percent is above the 15 percent maximum for this quantity.
5	E-beam review needs an inspector at the customer site to supply defect locations.
6	Fleets of 3 or more tools require Klarity defect data management.
7	Premium service requires a term of at least 2 years; 1 year was proposed.
8	Service terms run from 1 to 5 years; 6 years was proposed.
9	Stated fleet throughput does not match: the agent stated 200 wafers per hour; the validator computes 180.
10	Stated total does not match: the agent stated $11,047,800.00; the validator computes $11,042,800.00.
11	The 29xx broadband patterned inspector supports process nodes of 10 nm and above; the customer process is 5 nm.
12	The 39xx broadband plasma patterned inspector lead time is about 40 weeks; the customer needs it in 30.
13	The Surfscan SP7 unpatterned inspector is export-controlled and cannot ship to a restricted destination.
14	The destination is export-restricted; trade-compliance license review is required before this order can ship.
15	The high-sensitivity package applies to inspectors only, not to the Archer overlay metrology.