Cracking the Code: How to Sell AI Software in a Complex, High-Stakes Market
Master the Art of Navigating Complexity, Customization, and Selling AI at Scale
After nearly a decade of selling AI, I can tell you this: Yes, you can sell AI! But the real answer is more complicated: Yes, but it’s hard — because AI doesn’t fit into cookie-cutter sales strategies.
The reality? AI projects have an 85% failure rate. Whether they fail to deliver or collapse altogether, that statistic is staggering. It’s no surprise corporations are hesitant to invest millions in your AI product. Add to that the fact that few people in most organizations know how to implement AI, and you’ll find yourself up against strong headwinds.
I once worked with an account executive dazzled by the “upside in AI.” He thought it was a goldmine. Unfortunately, he didn’t sell anything for several quarters, and management let him go. Nearly every salesperson — including myself — underestimates how complex and lengthy the AI sales cycle can be.
Most of this stems from a basic misunderstanding: AI (whether machine learning, deep learning, etc.) is a tool to optimize a process. But it’s not a one-size-fits-all solution, and therein lies the challenge.
The Problem with Selling AI Like a Widget
Businesses, like people, love processes, and sales teams are no exception. Every salesperson I’ve ever worked with has their own sales process — sometimes tied to a broader go-to-market strategy, sometimes just based on whatever worked for the latest Chief Revenue Officer.
Books and research reports are written on sales methodologies — solution selling, value selling, the next buzzword — until someone makes millions and writes a book about their process, becoming the next Tony Robbins.
At its core, though, sales is simple: exchanging goods or services for a price. Your company builds AI software; Company XYZ wants AI to solve a problem. If your sales team closes the deal, upper management pressures them to shorten the sales cycle and sell more. The result? More sales processes.
The problem? Commoditizing AI software like a widget doesn’t work. AI is complex, often requiring significant integration, as seen in the case of C3.ai, which faced accusations of inflating revenue by passing off consulting services as software sales.
Selling AI as a service is tough. Yes, some companies succeed, but they’re often selling services powered by AI, not AI itself. AI tools — AutoML, deep learning models, feature stores — are like hammers and saws. They aren’t off-the-shelf products. Every AI model must be trained, tested, and tuned on unique data. Customization is inevitable.
This custom approach improves your chances of success, but it doesn’t guarantee it — and it frustrates sales teams who crave repeatability. At first glance, it seems like AI sales can’t scale. Or can they?
The Open Source Opportunity — and Trap
Open source is fantastic. Many of us use open-source tools daily, from Python libraries to feature stores. However, open source isn’t free. There’s a hidden cost in terms of time, effort, and resources.
Often, companies exploit open source, benefiting from the work of developers without contributing back. I’ve written about open-source exploitation before, where projects get used without reciprocation. So, how do you make money selling AI — whether it’s open source-based or not?
The Red Hat model provides an intriguing blueprint. Red Hat took the open-source Linux kernel and built an enterprise product around it. They didn’t just IPO — they became a center of gravity in the Linux ecosystem. Their secret? They gave away some products for free, while building irresistible enterprise versions that businesses couldn’t ignore.
If you’re in the AI space and want to sell an open-source product, take note: The backbone matters. Your open-source backbone must be strategic. It’s the foundation upon which future revenue is built.
How to Sell AI: Service vs. Enterprise
There are two ways to monetize an AI backbone. You can sell its output as a service, or you can build enterprise-grade capabilities around it.
Selling a service powered by AI is a common approach. Startups can offer subscription-based access to custom AI models or insights. This approach works, especially for smaller businesses, but it doesn’t usually result in multimillion-dollar deals. If you want those, you’ll need to sell to Fortune 1000 enterprises and offer something they can’t get for free.
Selling the enterprise version of your AI backbone is where the big money is. But it’s also harder. You’ll be competing with your free open-source version, so the key is to offer something compelling enough for enterprises to pay for it. That might mean building an entire ecosystem of tools around your AI backbone.
The challenge here is figuring out what you give away for free and what you charge for. If you give away too much, enterprises will take advantage of it without paying. Pivoting away from your open source license too abruptly could alienate your community — unless they feel the project is being exploited.
The Moral of the Story: Yes, You Can Sell AI
Selling AI isn’t just about having a slick sales process. It’s about having the right product strategy. AI sales success depends on building the right foundation — whether that’s with open source or proprietary technology — and understanding what your market needs.
You can sell AI, but you’ll need to rethink how it’s done. Forget cookie-cutter methods. The real work happens before the sales team even picks up the phone.