Are you 'Forming', 'Storming', 'Norming', or 'Performing' when it comes to AI usage in your team?

That's exactly how it felt when AI began showing up in our workflows.

Fortunately, in my team, there were already a few early adopters who were well-used to working with LLMs. The next stage was to make sure the rest of the group was comfortable and getting 'hands-on' experience using it in their everyday marketing tasks too.

I sought a training programme that gave a no-frills guide to using AI tools for marketing, and ended up taking a practical, interactive course from MMC Learning (for anyone who's interested and would like to learn more for themselves).

In order for my team to do the training, it was important to me that I do it first, for my own knowledge and development as well. I didn't want to ask the team to spend time on something I hadn't taken seriously myself, and I wanted to understand the limitations as well as the potential before encouraging wider use. Doing it first also meant I could answer questions honestly, based on my own experiences and learnings, rather than theory.

I was excited to see how a marketing-focused course would teach me how to make the most out of the AI tools at our disposal. Before doing the course, I thought I knew what AI and LLMs were capable of. How wrong I was, because suddenly a whole plethora of new possibilities were revealed before my very eyes. What surprised me most wasn't just how much they could do, but how easily misunderstandings could creep in if you didn't pause, question, and test what was being produced. It made me realise how much context, nuance, and judgement still matter.

And we're not alone in that gap between usage and guidance. In a recent workplace survey, roughly three-quarters of people reported using AI regularly at work, but only about a third had ever had any formal training on how to use it well. That mismatch feels familiar: lots of activity, but not much clarity on how to apply it confidently and responsibly.

When AI enters the picture, it doesn't really feel any different. There's curiosity, hesitation, and a lot of quiet experimentation happening in the background. At this stage, the goal isn't speed or output, it's familiarity. Giving people the space to explore, make mistakes, and build a basic understanding of how AI might support their work, without anyone feeling rushed or judged.

Some folks loved it straight away and were trying to see how it could fit into every process they could think of. Others asked thoughtful questions about when it made sense to use it and when it didn't. There were some who felt uncomfortable leaning on something that occasionally gave confidently wrong answers. It reminded me of those early days in a new team when people are all figuring out how much they trust each other, what the unwritten rules are, and where responsibilities really lie.

For myself, I found it incredibly difficult to create social media posts that I felt authentic and weren't too 'same same'. Yes, you know the one on LinkedIn where there are those usual culprits; starts with a shocking question or one-liner to grab your attention, then has an emoji list telling you why you're wrong. For content production, I do strongly believe in a human-first approach, as opposed to AI-first. After battling multiple times with AI to help 'make this post sound more human', I gave up.

There was this moment, early in a group session, where someone jokingly said, "Who's the AI whisperer now?". At first it was funny, but it actually flagged something important: a shift in dynamics. For some, AI was already a tool they reached for first; for others, it was something to be questioned and checked before anything else. That's classic Storming energy: different approaches bumping up against each other, old habits meeting new tech.

In that stage, leadership isn't about telling everyone what to do. It's about listening, acknowledging concerns, and helping people make sense of the experience. With AI, that meant talking openly about what it is good at, and just as importantly, what it isn't. Sometimes that was about accuracy or nuance. Other times, it was about context or ethics, like deciding when an idea generated by a model needs human thought layered on top.

What mattered most to me though, was creating space for curiosity, for mistakes, and for the little moments where a team member would pause, question, and discover something new about the tools they were using (and perhaps, more importantly, about themselves).

One real-world illustration from this phase came from some recent website improvements our team had been working on. Rather than describing a proposed module change in words, or asking a designer or developer to spend time creating a full mock-up, they used AI tools to quickly generate realistic, responsive prototypes in minutes. This helped validate whether an idea was worth progressing and gave the team something tangible to respond to. Conversations shifted from abstract feedback to practical iteration, making small website updates faster and more productive.

Another practical example from our wider team illustrates this perfectly. Instead of manually creating detailed demo projects in Jira to mimic real-world use cases (complete with pre-populated work items), AI was used to generate a realistic project foundation in a fraction of the time. From there, the team could layer in a bolt-on app and demonstrate capability in a meaningful context. This saved hours of setup and let the team focus on applying and testing AI in ways that actually added value.

This really felt like how a team settles into roles when a new person joins. At first, you're polite, then you might argue a bit, then you figure out who's good at what.

That shift didn't happen overnight, and it wasn't always smooth, but the change was visible in how conversations started to sound less like "Is this right?" and more like "What's the best way to use this here?".

We reached a point where we could say, "This is where it helps, but this bit still needs our thinking", and other people would nod because they'd seen the same thing. That felt like the team genuinely learning with the tool, not against it.

My own 'aha, this is actually quite good', was when I was conducting some market research on app marketplace ecosystems. I found a tool called Manus incredibly mindblowing. I fed it some instructions to research the HubSpot app marketplace, and it pretty much started talking back to me as if I were speaking to an enthusiastic graduate looking to make a stellar impression. After answering some of its follow-up questions, I let it conduct the research for 45 minutes, and it produced several documents with detailed statistics, graphs, explanations, you name it, about the HubSpot apps marketplace and also comparing it to other app marketplaces such as Atlassian and monday.com. I was astounded to see how detailed some of the deliverables were; if this were a human, it would have taken them weeks, if not months, to produce the same set of deliverables.

Again, think of that new team member who, after a few months, isn't someone you're figuring out any more, they're someone you rely on. You know their strengths, you know their quirks, and you know what they need from you to do their best work.

Research suggests that building trust with a new colleague can take several months of shared work, consistent interactions, and proving reliability. With AI, the same principle applies: you need time to experiment, test, and iterate before you can confidently integrate a tool into your workflow. It's not about letting the tool take over; it's about letting people use it intentionally, so their individual (and more importantly, human) skills can amplify what the technology can do.

That's when good things start to happen. Brainstorming might start with an AI‑generated draft, but the ideas, insights, and strategic direction always come from the team. A marketing brief might get a first pass from a model, but the real value comes when humans apply judgment and context to make it sing.

I find myself using AI regularly to help me with more personal matters, such as personal finance. Comparing rates, packages, and providers. It gives me hints and tips that I had not come across before, and where I often had to fact-check previously, I now see myself trusting it more often than not (I'd still do some fact-checking just to make sure, as nobody wants to make personal finance decisions based solely on AI recommendations now, do we?).

For work, where I see real value is helping us to create templates and processes—streamlined processes—to help my team get their work done more efficiently and effectively. It's not perfect, but I don't think it ever will be perfect, because not only are AI tools evolving so swiftly, but we as humans are also looking to keep up with the latest and greatest. And there's always another new feature or tool that gets introduced every week, which can be overwhelming even for the most AI-literate of us, let alone beginners.