Customer case

From 3 Days to 3 Hours: Onboarding a Configurable Product Catalogue with Claude

Kingspan Light + Air, part of the Kingspan Group, makes daylighting, smoke ventilation, and natural fire safety systems for buildings. The company had invested in Salesforce RCA to give its sales teams a fast, self-service way to configure and quote its products.

The Complexity Bottleneck in Product Catalogues

The catalogue itself was the bottleneck. Many of Kingspan's products aren't simple SKUs, they're highly configurable systems with dozens of attributes, hundreds of valid size combinations, and calculated outputs (area, load, CO₂ output, U-value) that all have to respect the underlying engineering rules.

Getting one of these products properly modeled in RCA meant translating a product engineer's Excel spec into Salesforce attributes, picklists, and CML (Configuration Modeling Language) constraint rules, the logic that stops a sales rep from quoting a combination that can't physically be built. That work needed a rare mix of skills: deep RCA and CML expertise, plus real familiarity with the product itself. Only a handful of people could do it well, and even for them, one product took 24 to 35 hours of focused effort. With a catalogue running to dozens of products, Kingspan's RCA investment was returning slower than it should have.

Overcoming the High Cost of Manual CPQ Onboarding

The math was working against the business case. At three working days per product, expanding the RCA catalogue fast enough to matter meant either tying up scarce senior specialists for months, or accepting that most of the product range would stay off the platform. This meant sales reps falling back on manual quoting for exactly the products where mistakes were most expensive. A single missed constraint rule in a configurable roof ventilator isn't a cosmetic bug; it's a quote for something that can't be manufactured.

Automating Translation with Claude and Human Verification

Nextview Consulting brought Claude into the onboarding workflow to take over the mechanical, high-volume parts of the job, the work that ate the specialists' time without needing their judgment.

Using Claude's code execution, Nextview had it read the source Excel specification directly, then generate the Salesforce-ready outputs from it: the full set of product attributes and picklist values, the CSVs formatted for load into RCA, and the hardest part, the CML constraint model itself, built from the same engineering logic in the spec sheet.

Nextview and Kingspan kept ownership of the parts that actually needed human judgment: validating that the generated business rules matched reality, and signing off before anything went live. Claude handles the translation work; people handle the verification and that division of labor is now how Kingspan onboards every new product into RCA.

The first product run through this production workflow was Kingspan's ECO Roof Ventilator: 20 attributes, 14 widths by 22 lengths for 308 valid size combinations, 51 control types, and 35 thermal-release options, 213 picklist values and 33 CML rules in total.

Accelerating Speed-to-Market and Eliminating Errors

Onboarding that product dropped from 24-35 hours to about 3 hours, roughly 90% faster. Cost dropped from €3,000-5,000 in specialist time to about €20 in AI compute plus review, a 99% reduction. Just as important: the 213 picklist values and 308 size combinations were generated and cross-checked programmatically instead of typed in by hand, cutting out the manual-entry errors that are easy to make and hard to catch in a rule set that size.

This is now Kingspan's standard process for onboarding products into RCA. 6 products, 30% of the catalogue have gone live this way since the ECO Roof Ventilator, with the same time and cost savings holding across products of similar complexity.