AI agents will be a threat to both traditional enterprises and big SaaS players

The greatest risk to enterprise companies will be small companies that leverage AI agents to act larger than they are.

I’m sensitive to criticisms, like those of Ed Zitron, that costs pose a threat to AI’s real-world value. However, this perspective still views LLMs as tools that create copy or code to support existing workflows.

As Ben Thompson points out, AI agents that can be trusted to complete tasks autonomously will, in the long run, be less expensive than humans—factoring in not only salary and benefits but also the costs of coordination and motivation (i.e., middle managers). This trend is already emerging.

He argues that enterprise companies will struggle to leverage AI agents due to their lack of data readiness and the breadth of undocumented tacit knowledge.

Meanwhile, SaaS companies—which make money by delivering tools to a growing user base of individual human workers with long-term customer value—will eventually find themselves unable to out-scale their customer acquisition costs.

It will be new-paradigm companies, built from the ground up with a few high-leverage individuals directing teams of AI agents, that will outcompete both classes of incumbents.

This is already happening with micro-SaaS apps, which leverage no-code tools to create sustainable one-person businesses, and with professional services agencies starting from scratch with minimal delivery staff.

The trend of increasing ARR per FTE will only accelerate as these agent-driven companies compete with established players.

Instead of adding an ultimately unhelpful “magic AI sparkle” to an existing SaaS experience, smart companies like Salesforce recognize the importance of agents going forward.