Thoughts on SaaS, AI, Agents

Lately I've been thinking a lot about the impact of AI and agents on SaaS and software products in general.

The market looks scared of what will happen to SaaS, e.g. Constellation Software, which invests in vertical market software, has experienced a significant drawdown.

Constellation has some idiosyncratic factors like Mark Leonard leaving and the question of whether its acquisition strategy can scale given its current size.

But as the CEO of Microsoft, Satya Nadella, said recently,

"I think the notion that business applications exist, that’s probably where they’ll all collapse, right in the agent era, because if you think about it, right, they are essentially CRUD databases with a bunch of business logic. The business logic is all going to these agents, and these agents are going to be multi repo CRUD, right? So they’re not going to discriminate between what the back end is. They’re going to update multiple databases, and all the logic will be in the AI tier, so to speak. And once the AI tier becomes the place where all the logic is, then people will start replacing the back ends, right?"

Not exactly encouraging. Indeed a lot of generic SaaS will likely perish.

But a significant portion of SaaS and software products do not just do CRUD operations, as Satya himself notes, they also implement some business logic.

For some of these applications, their business logic is a digital form of decades of domain expertise, lessons learned through interactions with clients, proprietary data sets and trial and error product iterations. All this is not something that's simple to create from scratch or teach an agent.

There's also a need for determinism and precision in many business lines, an AI hallucination could cause financial losses, non-compliance or reputational damages. It's difficult to see software being replaced in mission-critical tasks.

What I see currently happening in the market is a lot of software vendors are creating APIs for their products, so that an AI agent could execute workflows. The AI agents are becoming embedded into the software. This happens across the board from accounting software, e.g. Sage, to PTC's Creo.

In the future it will likely be a requirement for any complex piece of software to integrate AI, it will be a standard feature that users will simply expect to be there, like they expect a graphical user interface today instead of a command prompt. Whoever ignores AI will soon be "legacy software".

Given the above, one can form a checklist for SaaS and software investments in the AI era,

• Is the reason the product is sought after the deep domain expertise of its creators?
• Does the company work with lots of proprietary data that allow it to continuously improve the product? - a flywheel
• Does the product need to function deterministically to support critical operations and meet tight deadlines?
• Is the product used in a line of business that can lead to financial losses, non-compliance or reputational loss?
• Are the product owners heavily investing in embedding AI in their product?

One may also want to look into what positions the company is hiring for. Are they hiring Architects, ML engineers to modernise their product for the AI era?

Beyond positions in CSU and TOI, I hold smaller satellite positions in Autodesk, Nemetschek, Synopsis, CDNS and PTC Inc.