AI code generation, agentic coding, the hype train does not stop. LinkedIn is full of people who preach "adopt or die" and nothing and no one is even remotely fast enough to keep up with the latest developments.
In the meantime, leaders, be it first- or second-line managers, struggle with the most basic adoption problems. I think this is mostly driven by mindset and opportunity. Some companies won't enable their engineering teams due to legal hurdles and that's fine. It might change over time when we learn more. So, that's lack of opportunity.
I think there are always a couple of different perspectives converging on a manager. Senior leadership wants to increase productivity, some ICs have fully embraced it, most of the ICs are cautious. Engineers have been trained for years to build applications that fail gracefully, to be risk averse, and to be on top of everything. Knowing the code made people feel powerful and in control - and having written the code yourself made it even more memorizable.
Coding with an AI or letting an AI write code autonomously feels like giving up control and power at first. It's changing the interface between the code and the coder. A layer of abstraction, an indirection. On top of that, the first results are usually awful and feed into a subconscious confirmation bias that leads to rejection.
So, how do we as leaders go about this? How do we help our organizations change? AI is here to stay and we won't change that by rejecting it. How deep the transformation will go is not yet clear. AI enables faster iterations at a larger scale than ever before. This is why it feels so overwhelming. Organizations can't move that fast, especially if the business itself is already fast-moving. There is a limit to what an organization can adapt to.
I think the first and most fundamental change is the mindset of the organization. Curiosity instead of rejection, clarity and enablement instead of shadow-IT and secrecy. You will need to make one bet on one tool to start. Whether it's Claude or Gemini or whatever else is not really relevant to the organizational transformation itself. Your senior leaders, your procurement, your environment will have some opinion about it - even if you think differently. As long as it's not the garage side project of your CPO's nephew, you will end up with a usable tool that'll help you to lift your transformation off the ground.
The next step is to get your people excited about it. You can either build an internal demo, enable pull request reviews, or start a hackathon. Anything that forces people to use it beyond the initial "meh" works. Select one or two ambassadors that are intrinsically motivated to help and spread the word (and skills). Book trainings with the company that sells you the AI. Some of these things are budget-friendly, some of them need external budget. Either way, you can get started.
Fighting the inertia will be your task after a couple of weeks passed and the excitement goes down. Maybe you have a big release and the tool proficiency is not yet at a level where AI can help. This will kill your momentum and you will need to manage this proactively. Ask people how they use it for features or refactoring, make a show-and-tell event for 2-3 features where it really helped and make sure people learn from each other.
The LinkedIn crowd will keep preaching "adopt or die." The reality is messier: adopt deliberately, fight the inertia, and bring your people along. That's the actual job.