The Agentic Divide Quantified

The data indicates that the B2B software market has split in two over the past 18 months with the arrival of agentic AI. On one side: organizations that crossed into operational agentic AI. On the other: those still planning, considering, or watching. Wave 3 puts a number on the divide and on how fast it formed.

We believe crossing this divide is a necessity to remain relevant in the B2B world in the age of AI. That said, it is a statement of intention, not a guaranteed outcome. In the Wave 3 data, we suggest that crossing this divide means you are in the race; graduating to Runner maturity signifies you are ahead. We cover Runner maturity in Chapter 3 — but first, the superset majority that crossed the agentic divide.

This adoption rate is one of the fastest we have seen in B2B technology. Depending on the source and how you define start and end points, previous major technology waves — the internet, mobile, big data, cloud — took roughly a decade on average to reach similar adoption rates. Reaching majority adoption of AI agents in the B2B world less than four years after the launch of ChatGPT is, in our opinion, remarkable — and it is the obvious starting point for crossing the agentic divide.

Enterprises historically tend to lag their growth-stage B2B peers when it comes to technology adoption. For this reason, enterprise adoption is often a barometer of success for a new company, product, or technology trend. We believe that Enterprise-segment Tech decision makers’ preference for agents over SaaS software is notable — a statement that these enterprises intend to cross the agentic divide at, or ahead of, their growth-stage competitors.

Acceptance of AI autonomy increased significantly in the same nine months that agentic AI implementation doubled. We think this makes sense: organizations that deployed agentic AI updated their comfort with autonomous execution based on what they experienced.

But there is a function-level split. Only 33% of GTM decision makers who have crossed the agentic divide report trusting agents at some level of autonomy; Tech decision makers are nearly 2× that at 62%. The two sides of the divide are not just at different deployment stages — they hold different operating assumptions about how much autonomy they are comfortable delegating to AI agents. There appears to be a trust leap that needs to take place before a company — or even different functions within the same company — can permit an agent to function autonomously.

This is where the agentic divide becomes operational, not just attitudinal. Both maturity tiers report high agentic intention, but the gap in what each segment has actually shipped is another way to quantify the divide: 62% of Runners are operating autonomous agents in production workflows versus only 13% of Walkers. Closing this gap requires infrastructure, organizational practice, and time. Chapter 3, What Makes a Runner, may provide a blueprint for others to follow when it comes to closing the agentic AI intention gap.

The agentic AI adoption holdout position has nearly disappeared. In nine months, the share of decision makers with no plans to implement agentic AI fell 28 points. Combined with the implementation gain in Finding 1, essentially the entire “wait and see” population has moved — either into active implementation or committed planning for implementation. For organizations on the sideline, the data suggests the window for deliberation has effectively closed.

From Wave 4 onward, we won’t be talking about those who plan versus don’t plan to execute an agentic effort. Instead, we anticipate surveying decision makers on progress and results.