How a water-treatment equipment company replaced copy-paste quoting with a CPQ they built — and own
Their quotes were vendor PDFs copy-pasted into Excel, priced from one expert's knowledge. How a 10-rep industrial sales team built its own AI-assisted CPQ — now running 109 real proposals they own.

Manual quoting breaks the same way everywhere: the pricing knowledge lives in one expert's head and a copy-paste workflow everyone else improvises around. Quotes queue behind that person, the small line items get missed, and no two reps quote the same job the same way. This is the story of a company that fixed it — not by buying a CPQ suite and bending to it, but by building their own.
A real Customware build. Company details anonymized; every quote is verbatim from the people who built and use the system.
The quick story: custom CPQ case study (2026)
A Midwest industrial water-treatment equipment company — about ten sales reps and engineers quoting everything from filter housings to full RO systems — ran quoting as copy-paste: take the vendor's custom quote, paste it into their Excel format, and work up the price by hand, hoping no accessory got missed. Instead of licensing an off-the-shelf CPQ, their sales director described how she actually quotes — in plain language, session by session — and Customware's AI agents built that logic into a production quoting system the company owns outright. Six months from kickoff, the live app holds 109 real proposals ("those are not tests"), all ten reps are on it, and the pricing knowledge that lived in one expert's head is now software the whole team runs. No per-seat licenses, no vendor roadmap.
Why they wanted to own it
They believed what we believe: the people who run a business should own the software that runs it. Their quoting process WAS the business — which tank is "the one we want," which accessories a storage vessel always needs, who's responsible for what on a scope of work. That's company IP. Renting a generic suite meant translating it into someone else's model and paying rent on it forever. They wanted software that fit them, and belonged to them.
The challenge — quoting by hand didn't scale
"This is our vendor quote… we'll just copy and paste this whole quote into our format and then work up the price for it so we don't miss any details." — Director of Industrial Sales
Every quote assembled a deep multi-category catalog — softeners, carbon filters, RO systems, storage vessels with their mandatory accessories, pumps and panels, DI and UV, filtration down to the micron — across three different quote types (capital, preventive-maintenance, parts/repair). The symptoms were the same three we hear in nearly every engagement:
- Slow, queued, urgent. As the owner of a gasket manufacturer in the same program put it: "whenever we want to quote something, we want it today. Date required. Today. Yesterday… it's like the place is on fire."
- Leaky. "A lot of times those little accessories are what gets missed on our jobs" — the small mandatory line items were the margin leak.
- Inconsistent and trapped. "If you ask three different reps to quote [a 500-gallon cone-bottom tank], they might quote three different… tanks." Nine or ten reps, each self-managing, on knowledge concentrated in one director's head.
Why the usual fixes didn't fit
- Off-the-shelf CPQ suites — too heavy, too admin-dependent; the business adapts to the vendor's model, and you still don't own it.
- Consultants / hiring developers — quarters of waiting, and the knowledge ends up in the contractor's heads instead of yours.
- Cheap AI app-builders — this team had actually tried one first (their early quoting tool started life on a prototype builder — a pattern we see repeatedly; another client came to us migrating a Replit-built quoting tool). Prototypes demo well, then hit the wall on multi-user scaling and structural integrity. A quoting system carrying real revenue needs what's underneath: a real database, tested pricing logic, approvals — an owned asset, not a demo.
What they built (and what it changed)
Nobody on their side wrote code. The director described how she quotes, and a team of AI agents built it — "you're just asking the AI and it's going to do the work for you… it's called shaping the tool." The value landed in three places:
- Nomenclature: one product language, encoded once. The 500-gallon tank has one answer now ("this is it, this is the one we want"), customer accounts stopped duplicating, and quote conventions are standardized — the same fix a gasket manufacturer in the program made by locking free-text fields into dropdowns "so people aren't putting in different semantics of the same thing."
- SOPs in the system, not a document: storage vessels prompt for their mandatory accessories; scope-of-work responsibility splits are structured in the quote instead of remembered.
- Knowledge transfer: the "cheat sheet" a senior engineer kept for matching filter housings to filters became product logic — "it helps a new sales rep that has no idea." The director reviews instead of typing.
The first working versions came fast — usable software to test within the early weeks, on a Monday/Friday rhythm ("it's gonna be slow for the first couple weeks… then we're gonna set you up for success"). Building out the full system — all three quote types, the whole catalog, and every rep onboarded — reached stable production over about six months of those weekly sessions.
Where it stands
- 109 real proposals live in the system — "every one of those quotes you see in the live app are actual quotes of ours… those are not tests."
- All ~10 reps and engineers quoting in one tool, on one catalog, with one nomenclature.
- Pull, not push: the team is "very eager to have it up and running where they can just submit orders through it" — adoption is the wanting-more kind.
- And they keep extending it themselves — inventory and location management came next, built the same way.
"This makes me very happy actually… reminds me of how far I actually got in this." — Director of Industrial Sales, in a working session on her system
The part they mention first
It isn't the speed. It's that the system is theirs — built from their knowledge, shaped to their process, owned like the rest of the business. When the process changes, they change the software; nobody waits on a vendor. Or as another Customware operator — a drywall-company owner who built her own take-off quoting tool — put it in a one-line email, subject and all:
"I DID IT!!!!!!"
That's the feeling this is for.
Want to see what you'd build? Book a working session — describe how you quote today, and we'll show you what your version looks like. Related: Quoting software, built around you · Pricing
Frequently asked questions
Is this a real customer?
Yes — a live Customware system in daily use, holding 109 real proposals as of July 2026. Company details are anonymized for confidentiality; every quote in this story is verbatim from the people who built and use it.
How long did a system like this take to build?
The first working versions came within the early weeks — usable software to test right away. Building out the full system (all three quote types, the whole catalog, every rep onboarded) reached stable production over about six months of weekly working sessions. Simpler single workflows go live far faster; a deep, multi-type CPQ like this is the longer end.
What does something like this cost compared to buying a CPQ?
Customware is priced by implementation, not by seat — starting free for your first tasks, then from $2,500/month for a production build — so cost doesn't grow as your team does. The bigger difference is what you hold at the end: a system you own and can extend, instead of a per-seat subscription that ends when you stop paying.
Ready to fix this in your business?
Customware lets your team build production-grade software around how you actually work — by directing AI agents, not hiring a dev team or a long consulting engagement. Request early access.
