Share with your CRO
Threekit is betting that the biggest friction point in complex B2B manufacturing sales isn’t the CPQ system itself but the guided selling gap that sits in front of it. Its newly launched AI Sales Agent ingests unstructured inputs, things like voice memos, RFPs, and competitive quotes, and drives toward a valid configured proposal without requiring a complete spec upfront. The company reports 229% revenue growth for the agent over the past year, 10 million configuration-to-quote sessions monthly, and 150 enterprise customers, giving the claim some operational weight.
What this means for your business
If your sales org sells configured or engineered-to-order products, the number worth sitting with is “seconds rather than hours” for a first proposal. That’s not a UX improvement, it’s a rep capacity multiplier. The companies most exposed to this shift are those running channel and dealer networks where product complexity has historically been a training and attrition problem: the dealer who can’t hold the whole catalog in their head is the one who loses deals to a simpler competitor. An agent that reasons over configuration rules changes that equation without a full retraining cycle.
The sharper claim Threekit is making, one that Threekit as a vendor naturally has every incentive to overstate, is that CPQ and ERP are necessary but insufficient because they assume a complete and valid input before they’ll produce an output. Most manufacturing CROs already know this is where deals stall or get spec’d wrong. The agent’s value proposition is that it handles ambiguous, mid-process inputs and flags conflicts before they become shipment errors. If that holds in production, it shifts the ROI conversation from “faster quoting” to “fewer costly order corrections downstream,” which is a different budget justification entirely.
The falsification condition here is integration depth. An agent that reasons brilliantly over product data but requires a clean, well-maintained product catalog to do it will reproduce exactly the problem it claims to solve, just one layer earlier. Watch whether Threekit’s 150 enterprise customers cluster in verticals with mature product data governance or whether the agent is genuinely tolerating the messy, inconsistent catalog reality most manufacturers actually have. That’s the question a renewal conversation in year two will answer.
Concept deep-dive: Guided selling
Guided selling is the process of helping a sales rep or buyer navigate a product catalog that’s too complex to browse freely, steering them toward a valid, compatible configuration through questions, constraints, and recommendations. Think of it as a knowledgeable sales engineer embedded in the quoting workflow. It exists because modern manufactured products can have millions of possible combinations, most of them invalid, and without guardrails a rep will either undersell or spec something that can’t be built.
Based on reporting from Threekit Launches AI Sales Agent for Manufacturers, originally published 2026-06-18 03:00:00.
