AI Adoption in Sales: Why Depth Matters More Than Headcount

Should account executives be building their own AI agents? The answer is an emphatic yes, but that is the easy question to settle. The far more interesting question is how deeply an account executive should understand the way these agents actually work, and how widely that understanding needs to spread across an entire sales organization. Adoption on its own is not the goal, because shallow adoption produces shallow results. What truly separates the strongest performers is the depth of their usage and their willingness to rebuild the way they work. The organization itself must also be ready to evolve as new tools emerge and creative disruption reshapes how the business operates.

I sit on a team of roughly 50 account executives at Amazon Business, and somewhere between 75 and 80 percent of them use AI in some form, which on paper looks like healthy adoption. The remainder use it mainly to polish their emails, if they touch it at all. The problem is that this usage is shallow, because within that majority only about ten percent are using AI anywhere near its real potential. That small group builds tools, creates workflows, and treats the technology as something to work with rather than something to simply query like a chatbot. They are measurably pulling ahead of everyone else, and the gap shows up clearly in the opportunities they close, the quantity of emails they send, the quality of their launches, and the length of their sales cycles.

The biggest shift in my own thinking arrived almost by accident. While studying for my AWS Cloud Practitioner certification, I learned that it is far better to access resources programmatically by giving instructions as text rather than clicking through a user interface. At the time I did not understand why it mattered, because I was not a coder and the programming side of it meant little to me; I simply memorized the concept and moved on. Even so, the phrase stayed lodged in my mind.

The idea finally clicked when we gained access to MCP servers through our internal AI tool. I could now reach nearly every resource the company had from a single prompt, asking the interface to pull information from Outlook, read or update Salesforce records, submit tickets, send emails, and surface what I needed without having to navigate multiple websites and sources of truth. Everything ran through one conversation, and the efficiency it created completely reshaped how I worked. There is a clear before and after for me, because my work went from being scattered across a dozen tools to flowing through one place, and I moved dramatically faster as a result. That is exactly why learning to access your resources programmatically is the single most important step any account executive can take.

Once that door opens, adoption tends to follow a familiar distribution across any team. There are people who never use the technology, steady adopters who rely on it for most tasks, and faster movers who embrace it early. Beyond them sits a small and almost unreasonable few, the tinkerers who build their own agents and constantly push the limits of what everyone else assumes is possible. These are the people an organization should watch most closely and engage with most actively.

When one of those tinkerers builds something genuinely new, the organization has to be ready to capitalize on it, and this is where most companies (including Amazon) fall short. We once had someone rapidly build a tool with AI that could have transformed the way customers access and interface with Amazon Business. The use case was strong, enterprise clients were ready to pay a premium to use it, and the return on investment was quantifiable, driven by added users and greater account stickiness. The problem was that there was no operational framework in place to recognize the new tool, assign a product manager and developers, and turn the idea into a finished product. Asking account executives to be creative is not enough on its own, since a genuine breakthrough demands a way to spot it, measure its impact, and act on it even when doing so means setting the original plan aside.

The conclusion I keep arriving at is that AI adoption in a sales organization is not something you simply extract from your account executives, because it has to move in two directions at once. From the bottom up, it begins with account executives learning to access their resources programmatically and completely rethinking their daily workflows. From the top down, it requires leaders who are willing to be disrupted, not only in their workflows but in their assumptions about how the business itself operates. Get both halves right, and the small group of power users stops being a set of outliers and instead becomes the standard everyone is measured against.