Why AI Is a Leadership Practice
Series note
This is Part 4 of our four-part Field Notes series on AI-era leadership. Across the series, we have shared what building with AI has taught us, how it changes the way a small studio works and what tools sit underneath our day-to-day practice. This final piece is the thread underneath all of it: AI works best when the leadership around it is clear.
Why AI Is a Leadership Practice
Note: this piece references our existing hero article on this topic. Need the URL before this goes live, marked below with [LINK NEEDED].
Here is the view we keep coming back to:
The leaders who work well with AI tend to be the ones who already understand how to set direction, give context, define quality, offer feedback and hold ownership.
That is why we see AI as a leadership practice.
We have written about this before on our site [LINK NEEDED], but it is worth saying again here, because it is the thread underneath this whole series. Every tool we have talked about, the Story Harvester, the inbox agent, Cowork, Notion, Riverside FM, only works as well as the person or team directing it.
AI adoption is rarely just a tooling challenge. It asks leaders to translate a new strategy into real behaviour.
What changes in meetings? What changes in handoffs? What decisions still need a human? What can be delegated? What does the team now own together?
That is leadership work.
Asking better questions
We do not really believe prompt engineering needs to be treated as a separate technical skill for most leaders.
Asking strong questions gets you a long way there. The same kind of questions you would ask before handing a project to a person:
What outcome are we actually going for?
What does good look like?
What context would make this easier to do well?
What constraints matter?
What should stay human?
Those questions are useful because they force clarity. They slow the brief down just enough for the work to move with more intention.
Making expectations specific
The leaders we work with already know this about people: vague expectations produce vague results.
The same is true with AI, only faster. Give it a fuzzy brief and you will get a fuzzy result back within seconds. That can be frustrating, but it can also be helpful. It shows you where your own thinking was less clear than you thought.
That same clarity gap shows up across teams too.
A team can understand the direction and still not know how to behave differently on Monday. They may know AI matters, or that the business is changing, or that new ways of working are needed. But unless the expected behaviours are specific, familiar patterns usually keep running the work.
Clarity is not a one-time announcement. It has to show up in decisions, meetings, handoffs, feedback and follow-through.
Giving feedback that improves the next version
Good managers give feedback that is specific enough to act on.
The same is true with AI. Telling it "this is not quite right" usually gets you another uneven draft. Telling it exactly what is off, what good looks like and what needs to change gives the next version a much better chance.
That is useful practice.
The leaders who are already skilled at this with people have a real head start. The ones who are still building that muscle have an opportunity to practise in a lower-stakes environment, with a collaborator that does not take feedback personally.
The point is not to become more mechanical. It is to become more precise, more aware of your standards and more able to communicate what quality means.
Distributing ownership
Ownership, trust and accountability are central to the way we think about leadership at Gather & Grow, and they do not stop applying because one of the collaborators is AI.
Someone still owns the outcome. Someone still needs to decide what good looks like. Someone still needs to review the work, challenge what comes back and make the final judgement.
At the team level, this becomes even more important.
If AI is introduced into a system where decisions already route back to the same few people, it may simply increase the volume of work those people have to review. If roles are unclear, AI can make that lack of clarity more visible. If people are waiting for permission, the tool will not automatically create ownership.
The leaders who will use AI well are the ones willing to look at the system around it.
Where is work getting stuck? Where does everything still route back to one person? Where are people waiting for reassurance or a clearer brief? Where is the tool exposing something that was already fragile?
Those signals are useful. They show us where more clarity, trust or shared ownership is needed.
Start with the work around the tool
If AI is creating friction in your team, the tool may not be the first place to look.
Start with the working conditions around it: expectations, decision rights, feedback loops, ownership and the behaviours people are being asked to practise.
That is where AI gets interesting for us.
It does not only help us produce more. It reveals how work actually moves, where clarity is missing and what leadership is being asked to hold.
If you want the fuller version of this idea, read the original piece here [LINK NEEDED].
And if you are trying to make sense of AI in your own team, our starting point would be this: look less at the tools for a moment, and look at the work around them.
Where does ownership sit?
What needs more clarity?
What behaviours need to change for the strategy to become real?
That is where the useful work begins.
Read the full series
If you are exploring how AI could support your own team, the earlier pieces in this series offer the practical context behind this point of view:
Part 1: What Building With AI Has Actually Taught Us
Part 2: How a Small Studio Uses AI Without Losing the Human Thread
