AI-Powered CPQ: Market Trends and the Competitive Landscape

We've watched this page's promised "2025–2028" window start burning down within months of publishing it, which tells you something about how fast AI in CPQ is actually moving. This is our most perishable page on the site, and we now refresh it every quarter rather than annually.

What is AI-powered CPQ?

AI-powered CPQ is ordinary Configure, Price, Quote software, a rules or constraint engine that guarantees a configuration is buildable and correctly priced, with an AI layer added on top. In practice that layer shows up in three places: a conversational front end that turns "what does the customer need" into a valid configuration without the salesperson clicking through every option; a recommendation or pricing-guidance layer that suggests options, bundles or discount boundaries; and, increasingly, an assistant that helps the people who build and maintain the underlying product model do that work faster. None of these three things touch the part of CPQ that actually matters most, the guarantee that what gets quoted can be built. That guarantee is still deterministic rule logic, and we don't expect that to change soon. See our full explainer of what CPQ is if you're new to the category.

Trend 1: conversational and generative quoting

The most visible change over the last two years is the front end. Instead of a form with dropdowns, several vendors now offer a chat-style interface: describe what the customer needs in a sentence, and the system proposes a starting configuration for the rules engine to validate and price. This is a real improvement for complex products with hundreds of options, where the bottleneck was often "which of these 40 questions do I even need to ask this customer." It is not a new pricing model and it is not a new validity guarantee, it's a faster way to arrive at the same valid, priced configuration a good CPQ already produced. Where this genuinely pays off is in guided selling for reps who don't know the product cold, or in dealer/self-service scenarios where the buyer doesn't know the vocabulary at all. We cover the underlying mechanics in our guided selling piece.

Trend 2: predictive pricing and deal analytics

The second trend is AI applied to the pricing layer, surfacing a recommended discount range, flagging a deal that looks likely to slip on margin, or highlighting which upsell historically closes with a given configuration. This is closer to traditional deal analytics than to generative AI, and it's the area where vendors with a strong pricing heritage, Conga's PROS B2B acquisition being the clearest example, have a structural advantage over configuration-first specialists. If your pain is price realization rather than configuration validity, this is the part of the AI story worth pressure-testing hardest. We go deeper on the pricing side in price waterfalls and CPQ.

Trend 3: self-service and auto-approval

AI-assisted configuration is also lowering the bar for genuine buyer self-service, letting a distributor or end customer configure and get a valid, priced quote without a salesperson in the loop at all, with approval routing kicking in automatically only when a deal falls outside normal bounds. This isn't hypothetical for us: our customer Swift Lifts already lets distributors configure and quote independently from a single unified model, replacing what used to be 40 separate spreadsheets. Adding AI on top mainly helps distributors who don't know the product as well as an OEM does, it shortens the learning curve, it doesn't remove the need for a correct underlying model. That model discipline came first, long before any AI layer was on the table.

Trend 4: modelers using AI, not just sellers

The least-discussed but, in our view, most durable trend is AI aimed at the people who build and maintain the CPQ model itself, not the salesperson. Several vendors, including Tacton, now ship assistants aimed at speeding up how a modeler drafts or extends configuration logic. This matters more than another chatbot for buyers, because the actual bottleneck on most CPQ programs isn't the sales interface, it's the ongoing work of keeping a large product model accurate as the product line changes. An AI that shortens that maintenance cycle is worth more to most manufacturers than one that makes the buyer-facing chat feel more natural.

The build-your-own wildcard

A trend worth naming honestly: some manufacturers, watching how good general-purpose AI coding tools have become, are experimenting with building a lightweight configurator themselves rather than buying a platform. We think this is a real and interesting development, not a fad to dismiss, and also that it inherits the same problem every home-grown rules engine has always had: it works fine until the product is complex enough that an unvalidated configuration slips through to production. We're tracking this "vibe-coded CPQ" question in more depth in CPQ for B2B manufacturing in 2026.

Where the major vendors stand on AI

Full disclosure before this table: cpq.se is a Tacton reseller and implementation partner for small and mid-sized manufacturers, and we advise enterprise organizations on vendor choice. We build with Tacton and know it best, evaluate the row below with that in mind, and always demo any vendor's AI on your own product rather than their slide deck.

[VERIFY: current public detail on ServiceNow/Logik.ai and Epicor CPQ AI feature names and release dates, for a future refresh of this table]

What AI changes in CPQ, and what it doesn't

Our take

Most AI-in-CPQ coverage reads like vendor marketing dressed up as trend analysis. Our practitioner view, after 25 years combined in CPQ projects: the sales-facing chat interface gets the headlines, but the AI that actually moves the needle on a manufacturing CPQ program is the kind aimed at modelers, because the model is where CPQ programs succeed or decay. If you're evaluating vendors on AI, ask less about the demo and more about which half of the problem, selling or modeling, their AI actually targets, then check that against where your own program is bottlenecked. For most product-complex manufacturers we work with, that bottleneck is still modeling capacity, not sales-floor conversation quality.

Next steps

See how the vendor field breaks down beyond AI in our comparison of the top CPQ vendors, or check what a real implementation costs on our CPQ pricing page, get in touch if you want a straight answer on whether a given vendor's AI claim holds up on your product.

Frequently asked questions