CPQ. A category so disliked that it has an industry nickname: “Causes People to Quit.” Yet, it’s so deeply embedded in how companies sell that its shortcomings have become accepted operating procedure.
Reps rely on Deal Desk to get a quote. A change in pricing requires someone who knows how to reconfigure the system. And when that person is unavailable, or the system can’t handle the deal, the sales cycle waits.
And time kills all deals.
We’ve reached a point where the software at the center of how a company sells should be some of its most capable. Instead, revenue teams have learned to accept its limitations. We’ve built processes around them, hired people to manage them, and treated the whole thing as the cost of doing business.
We figured it was high time to change that.
That’s why we built Roadrunner.
The team we needed to actually solve it
Roadrunner was born out of Kleiner Perkins, the firm’s fourth incubation in its 50-year history. The firm hosts a group of enterprise CIOs, and for two years, their number-one pain point was CPQ. Across multiple companies, it was the lowest-rated software in their NPS surveys. Absolutely insane.
AI had changed how companies thought about pricing, and their underlying revenue systems couldn’t keep up. Reps were stuck with slow, archaic systems built decades ago that could barely handle “swap 50 seats of Starter for 100 seats of Advanced.” There was no hope for a three-year ramped deal with a multi-commit drawdown model and a few free AI credits sprinkled on top.
Our co-founder and CEO, Joubin Mirzadegan, ran the CIO group as an operating partner at Kleiner Perkins. He’d spent years as a sales rep living that exact problem and couldn’t believe it had gone unsolved while market forces made it worse. And the cherry on top: the incumbent had stopped selling its CPQ solution, leaving thousands of enterprises figuring out where to go next.
He met our co-founders Ajay Natarajan, Chief Product Officer, and Eugene Shao, Chief Technology Officer, both second-time founders, former Caltech roommates, and Kleiner Perkins fellows, who had been chasing down the same CPQ problem. They had realized the issue lay at the data model layer, and slapping on a pretty AI wrapper wasn’t going to solve anything. Bad data model, bad outcomes.
And so Roadrunner was born.
The business is moving faster than the system
The quote is where the problem becomes visible. Underneath it is a model of what the company sells, how it prices, and what it will allow. Every new product, pricing structure, and negotiated term has to fit into that model somehow.
When it doesn’t fit, someone finds a way. Another SKU. Another rule. Another spreadsheet. Over time, the workarounds become part of the infrastructure, and the people who remember why they exist become responsible for keeping the whole thing running.
And the problem is getting harder. Companies that used to have simple seat-based pricing are moving toward usage or consumption-based models, often with credits, commitments, or a mix of subscriptions and usage. AI is forcing companies to reconsider what they charge for, how they measure value, and what their customers are actually buying.
A company should be able to work through those questions without its CPQ becoming a limiting factor. But if a new pricing model means months of configuration, companies end up selling what’s easiest to configure instead of what makes sense for their customers. You can build a product your customers want and still struggle to sell it the way they want to buy it.
The same technology creating this complexity gives us a way to manage it. But only if the systems underneath can represent the business as it actually operates, and keep up as it changes.
What we’re building
We built Roadrunner around a new data model, with agents doing the work of configuration and quoting. Products, pricing, policies, and customer agreements give those agents the context they need to act.
The interaction is straightforward. Describe the deal you’re trying to do. The system puts it together using the company’s rules and brings in the right people when a decision needs approval. Prompt. Quote. Approve.
Making that dependable is where the real work is. An agent can interpret what a rep means. The infrastructure has to make sure the pricing is calculated correctly, the terms are allowed, and the approvals actually happen. Your pricing policy isn’t a suggestion.
We’ve built alongside enterprise teams because this has to work with the deals they actually do. The amendment halfway through a contract. The customer on an old pricing plan. The exception someone approved last year. Those details are part of the job, and they have to be part of the system.
The knowledge that lives with RevOps and Deal Desk is essential to getting this right. Those teams understand the business. Their expertise should shape the rules the system operates by, so they don’t have to personally shepherd every ordinary quote through it.
Where this goes
We want a company’s ability to sell to keep pace with its ability to build.
Imagine launching a product or changing how you charge for it without immediately creating a backlog for the team that manages your revenue systems. Reps can get accurate quotes within the rules you’ve set. Deal Desk spends its time on the deals that need judgment. RevOps can change a policy, understand its effects, and trust the system to carry it out.
And that understanding should survive beyond the quote. What the customer bought. What was agreed. What changed along the way. When it’s time to expand or renew, that context should already be there. Nobody should have to piece the deal back together from a contract, a spreadsheet, and an old Slack thread.
That’s the future we’re building toward across quote-to-cash: systems that understand what’s been agreed with a customer and carry it through to the next transaction.
There will always be hard decisions about how to price, what terms to accept, and when to make an exception. Those decisions deserve people’s attention. The work of getting them reflected in the system should take a lot less of it.
We’ve spent long enough asking people to accommodate their software. It’s time the software caught up.
Come build with us
Entire companies run on software their teams can’t stand. We think that’s a hell of a problem to solve. If you do too, we’re hiring.
— Joubin, Ajay, and Eugene
