Roadrunner raises $27M from Founders Fund and Kleiner Perkins to modernize AI-era CPQ and revenue infrastructure

Roadrunner, an AI-native CPQ (configure, price, quote) platform incubated at Kleiner Perkins, has raised $27M across two rounds: a $5.2M seed from Kleiner Perkins and a $22M Series A led by Founders Fund, with Kleiner Perkins doubling down.

The problem

The company's origin is a CIO dinner circuit that Joubin Mirzadegan runs at Kleiner Perkins. The CIOs told him CPQ was the most broken software inside their organizations, in some cases ranking as the number one negative NPS tool in the business. Mirzadegan, who comes from a sales background, did a full market map, found nothing adequate, and built Roadrunner instead.

Why now

The incumbents are structurally exposed. Mirzadegan argues that a new CPQ vendor emerges roughly every decade because each major technology shift, from software to SaaS to AI, brings a new pricing model, and the legacy data models can't keep up. SaaS moved pricing to subscriptions; AI is moving it toward usage and consumption-based billing. If a sales rep can't get an accurate quote out the door under the new model, the existing platform is broken. Roadrunner is built to price those deals from the start.

“We incubated the company at Kleiner Perkins. CIOs told us it was the most broken problem inside their company — in some cases the number one negative NPS surveyed software inside their business. Every ten years there's a new technology shift, and AI is very likely going to be usage or consumption based billing. Every time that happens, you need somebody that can actually price those deals.”

The product and the technical bet

CPQ looks like a natural AI problem because the underlying inputs, price books, approval rules, volume-based discounting, are structured or semi-structured text that models reason over well. Roadrunner's architecture layers an agent at the top, a set of policy engines that enforce deterministic rules over a probabilistic AI system in the middle, and a flexible data model underneath designed to accommodate pricing models that don't exist yet. The durability of that bottom layer matters: if the data model can't handle tomorrow's billing construct, the whole stack breaks again in five years.

The harder challenge is operational reliability. Roadrunner sits in a production system where downtime means deals don't close and contracts may not get out the door. Mirzadegan frames that reliability constraint, getting enterprise-grade predictability out of an agentic architecture, as the genuinely hard problem the company is solving.

The sales motion

The two primary buyers are CIOs, responsible for enterprise software, and CROs, responsible for deals closing. But the actual buying process touches DealsDesk, RevOps, finance, sales, and IT simultaneously. That breadth makes it difficult to sell and difficult to build for, and it raises the additional hurdle of convincing large enterprises to run a mission-critical production system on an early-stage startup. Mirzadegan acknowledges all three as real friction. The upside of clearing them is that the product becomes deeply embedded once it lands.

Founders Fund and the Anthropic question

The sharpest investor question during the Series A was whether Anthropic or a similarly capable foundation model would simply absorb the problem. Mirzadegan's answer is that if Anthropic goes after enterprise CPQ directly, the category is unwinnable for everyone, and the rational move would be to hold Anthropic equity instead. His actual view is that CPQ is niche enough that frontier labs won't prioritize it, and specialized enough that the policy-engine and data-model layer represents durable differentiation even if the underlying models commoditize.

Founders Fund's Trey Stevens led the Series A.