I’m willing to wager that there’s a huge discrepancy between what you see SaaS companies doing, and what they’re discussing in their boardrooms.

What we can see: they’re bolting AI functionality onto everything that moves. Some of this works quite well; for example, I designed The Antidote using Figma’s AI agent, albeit with quite a lot of direction-setting required. Others work less effectively - I’ve had a number of battles with Atlassian’s AI agent, Rovo, that have driven me round the bend. But I digress.

What SaaS companies are actually terrified about: AI capabilities will make their platforms entirely obsolete. Why should customers pay to licence a platform when you can get AI to build it for a fraction of the cost?

If you don’t believe me, here are some clear examples of the direction of travel.

• A plucky developed called Brandon Thomas created free clones of the entire Adobe Suite and Microsoft office suite. Reviews have been quite positive. To be clear though; the risk isn’t that companies will jump across to use Brandon’s products; the risk, or likelihood, is that they’ll start building exactly what they need themselves.
• When building The Antidote, I needed a blog engine for the News and Views section. I also needed a Digital Asset Management system (like Adobe DAM), a customer support system (like Zendesk), a Marketing Campaign Management System (like Adobe Campaign), and more. I built it all myself, purely using AI. It took less than a week, including testing.
• One of my clients wanted to build a Help Centre - actually, four of them for different parts of their ecosystem, all sourced by a number of shared knowledge bases. We wanted AI driven chatbots to help people find their answers, and also draft responses to support enquiries for humans to review/amend and then send. After 3 months of banging my head against the wall trying to get Atlassian’s JSM and CSM products to do what we wanted, we threw in the towel - and I built the solution myself. It took me about 3 days, plus a day of developer time to integrate into their existing platform. It looks fantastic - much, much better than the Atlassian solution.

I don’t seriously expect a mass exodus away from existing SaaS platforms - the cost of change can be high. There’s re-skilling the workforce to consider. Data or customer migration may be involved. And for some platforms, aka marketplaces, the value is found in the well-established relationships that only exist on the current platform. Also, significant regulatory hurdles may have been cleared. Plus, if someone senior approved the licence, there’s often significant pressure to continue getting return on that investment.

But still, I think that companies will come to the realisation that - in an increasing number of cases - you can build exactly what you need, for a fraction of the cost of a SaaS licence. In the time you would’ve taken to complete the procurement process, it would already be built. And you would never again have to wait in line while your feature requests languish in their backlog. Talk about operational agility!

Smaller and newer companies will be better placed to capitalise on these opportunities. And all companies going through a procurement process now have new options to explore, and new leverage they can exercise. At the very least, it’s worthwhile to consider how your SaaS platforms sit in your digital ecosystem alongside AI agents and tools that you’ve built and control, and gradually shift the balance in-house where the opportunities make sense.