Every July, development teams do the same thing. They close out the fiscal year, pull the final numbers, and start building the annual fund plan for the cycle ahead. Goals get set. Timelines get mapped. Someone makes a calendar. The plan looks right because nothing is running yet, and when nothing is running, nothing is visibly broken.
You cannot run a marathon on your first day. You can put on the shoes and walk to the end of the street. That is the whole method.
If you know AI could help but have no idea where to start, you are not behind and you are not alone. It is one of the most common things a team runs into right now, and it is rarely about the tools.
If you bought the tool and somehow made more of a mess, you are not behind and you are not alone. It is one of the most common things a growing team runs into, and it is rarely the software’s fault.
If you cannot say what is in your pipeline right now, you are not behind and you are not alone. It is one of the most common things a development leader runs into, and it is rarely about the software.
The spreadsheet was doing two jobs at once: holding the friction and holding the knowledge. Automating the follow-up can oversaturate the people you were carefully tending. The wrong system does not just fail to help; it quietly erases the thing that was generating the growth.
Six months after the rollout, the work has routed back to inboxes because the platform was treated as the system when it is not. Who owns a piece of work, what done means, who owns the system itself are calls a leader has to make and then stand behind. None of that ships with the license.
There is a meeting on your calendar that you have moved twice, and a thinking partner set up correctly will not let you look away from it. The starter prompt that makes this work closes the back door: “You are not here to validate my thinking” cuts off what avoidance is looking for in the first place.
Ninety-two percent of nonprofits are using AI. Four percent have built a workflow. Eight people in the same organization can be using the same tool in eight different ways, with no shared prompt library and no internal agreement on what data goes into a model.
