The AI Strategy Divergence in SaaS
Headless or not, that is the question
In early February, I published an article titled “AI Disruption and the Software Bloodbath” that outlined an evaluation framework I was using to try to determine early winners and losers in the pre-AI era software landscape.
In it, I talked about software companies whose values were most tied to the user interface being the most at risk.
Along those lines, this is a short piece to update my framework, and it’s very simple.
I think SaaS companies that recognize the friction human users experience with their UI’s as a weakness and pivot to “headless” architecture and business models will be best positioned to win, and have already meaningfully moved in that direction- CRM and TEAM are noteworthy examples.
See CRM’s Headless 360 announcement, where all aspects of their products will now be exposed as API endpoints and MCP connectors to make consumption of their products as easy as possible for AI agents, reducing the need for humans to login to their platform. CRM also has a context advantage in that a ton of business context already lives in Slack.
TEAM’s MCP connector for Claude is the 7th most popular connector in a list that includes consumer-use connectors, illustrating the uptake of Anthropic’s AI into developers’ Jira and Confluence workflows. Another shoutout to Slack here, since many developers are using Slack as a command center for this. Once again, avoiding the need to login to the UIs of TEAM’s products.
I recently watched NOW’s analyst day presentation, and it seems they are taking a different approach. If you want to build, deploy, and manage AI agents into their workflows, for the most part, you’re still logging into their interface to do it.
NOW feels like a company leaning on the fact that so many cross-organizational workflows are built around their existing products that there isn’t much threat of disruption. They seem to expect agents will be built and deployed within their platform automatically.
CRM benefits from cross-organizational entanglement as well, but has (humbly?) acknowledged the shift to agents as primary interfaces is already happening and is adapting its entire business to that shift.
This early divergence in approach also signals a bet on whether or not the nature of work is truly changing. If CRM is saying, “Agentic AI will change the nature of work forever,” NOW seems to be saying, “Work is going to happen the same way it always has, but now there will also be AI agents in the mix.”
It’s too early to know which bet will pay off, and my AI Research bias at Marker is perhaps influencing my judgment, but I am in the camp that believes the nature of work is already changing in irreversible ways.
As a final thought, it’s my belief that organizations who recognize the need for the Human + AI Partnership - rather than replacing human employees with agents - are in the best long-term positions to win.
Along those lines, I’m interested in companies that are re-skilling their employees for the AI era. The ones firing employees and replacing them with AI agents are in for a rude awakening and are cutting off their noses to spite their faces.
Ironically, the ones taking time to re-skill employees and shift to headless are most likely to experience pain in the stock prices as they make the change, while those firing employees for operational gains may appear to win in the short term.
CY2026 is a transitional year where software companies are drawing their battle lines. 2027 is going to be very telling.