CPQ for custom products: when each fits.

Classic CPQ assumes a sales rep at a keyboard during business hours. The buyer asks the rep, the rep drives the guided forms, the CPQ produces the output. Poly9 AI assumes the buyer is asking in plain language on your website, on a showroom floor or at a trade show booth, and answers from your catalog. For custom products whose buyers do not wait for 9 a.m., that difference decides the sale.

Both configure and enforce compatibility. Classic CPQ does it through admin-maintained forms: dropdowns, conditionals, validation rules that break when the admin has not anticipated the combination. Poly9 AI enforces compatibility inside the cost recipes you write in plain language, so “washdown-rated motor without a stainless enclosure” is refused with the reason before a price appears. Ask for an option you have not priced and it says price on request.

This is the line between an honest quote and a hopeful one. Classic CPQ prices from rule tables, and when a rule is missing the quote fails or the rep rebuilds it in Excel with a guessed number. Poly9 AI prices from your cost recipes and grade matrices through one engine. When the rule is missing it returns price on request and drafts the request for your estimator. It never guesses.

The Excel rebuild is where CPQ value leaks for custom products. The CPQ produces a line-item document; your team exports it, fixes the formatting, adds the options the CPQ does not know, adjusts the guess and emails a PDF. That is how 43% of manufacturers end up quoting from spreadsheets (Tacton, State of Manufacturing 2025). Poly9 AI sends the quote itself: the buyer opens it on the train home, sees the product in the chosen options, approves in one tap and replies in the thread.

Presentation is outside classic CPQ; it lives in decks and separate tools. For custom products the render usually sells before the number does. Poly9 AI renders your product in the chosen options from your materials library, turns a mood board into a priced collection and hands the buyer one link. A cabinet dealer quotes nine windows and a door for a whole house in one collection. A fabric house shows eight colorways of a king duvet, each priced from its grade.

Here the CPQ wins. Delegated approval chains, credit checks and validated order entry into the ERP are what classic CPQ suites were built for, and Poly9 AI does not replace them. A Poly9 AI quote can require one internal approval before the buyer sees it, and the approved quote becomes an order in Projects. If your policy needs more steps than that inside the ERP, keep the CPQ for order entry and put Poly9 AI in front of it.

A CPQ knows what the admin programmed. Poly9 AI learns from every question, quote and collection: which buyer prefers grade F when the upcharge is small, that nailheads usually go with box pleats in your quotes, that one rep always forgets to ask about the wood species. The cost recipes are still yours and still enforced. The memory layer suggests what to ask next.

Per-seat pricing punishes companies whose quoting is done by designers, dealers and occasional reps. A showroom with twenty designers who each send two quotes a week pays for twenty seats, and most of them sit idle. Poly9 AI is one credit per sales action, so the cost follows the quotes. $100 buys a thousand credits, everyone on your team is included, and four reps sending 30 quotes a week use about 500 credits a month.

CPQ implementation is a project: weeks to months of discovery, rule migration, admin training, rollout and ERP integration. Poly9 AI is an account. Create it, upload your catalog or start with your industry’s sample catalog, write the first cost recipe and send the first certified quote in the same session. A 37,400-SKU fabric line with grade matrices loaded in one day once the files were in hand.

Questions about CPQ tools.