There is something the AI conversation keeps skipping over.
We talk about the concept. We talk about the tech. We talk about the models, the agents, the launches, the demos, the framing slides at every keynote. There is no shortage of words. What is in short supply is the much quieter conversation about adoption. The actual practice of using these tools, in your work, every day, with intent.
Adoption is the part that matters.
I want to write about that.
The gap that keeps coming up
Walk into any conference room, any meetup, any LinkedIn thread on AI right now, and you will find people deep in concept. The strengths of one model over another. The arrival of agents. The ethics, the regulation, the risks, the philosophical question of what intelligence even means now.
All of that matters. None of it tells you what changed in your work this week.
When I get into one-to-one conversations with customers, or with some of the practitioners I meet, the topic moves to a narrower question. Are you actually using these tools? In what? How often? What have you stopped doing because the agent is doing it? What have you learned about the seams between the human and the agent?
The answers are usually quieter and more honest. That is the conversation I want more of. That is where the real adoption decisions get made, not in the keynote halls.
A confession before I go any further
I want to be honest with you about something.
I think of myself as someone who stays at the front of the tech. Fifteen years in the Microsoft ecosystem. A Microsoft MVP. Daily conversations with the product teams and the MVP community. That is the position I am used to operating from.
A few months ago, I noticed I was not at the front of this one.
There are private chat groups I sit in. One with other MVPs. One with parts of the Microsoft product team. Both are firehoses. The MVPs are sharing new patterns daily. The product team is shipping faster than any of us can absorb. Most days I can keep pace. On AI, I noticed I was not. Concepts I had not got around to. Tools I had heard of but not used. New ways of working that other people had already internalised while I was still reading about them.
I was using AI. Just not in the way I wanted to be. I was leaning on it for the obvious things and missing the structural shift. And because of that, I was still spending my personal time on tasks that could have been automated or handed to an agent. Which left me with less time to focus on the new things, less time to shape my AI practice in the way I wanted. The gap was widening on me, not closing.
So I made a decision. I accepted, privately, that I was behind on this particular wave. Not behind on tech in general, but behind on the specific discipline of using AI deliberately. Then I put my head down and started.
Slowly, but steadily. I gave it a couple of hours a day. I picked things in my routine I could automate, set up the agents, watched them, corrected them, and let them take over the heavy lifting where they could. The space that opened up was the point. It freed me to focus on the work that genuinely needed my attention, which in turn let me catch up on the very concepts and patterns I had felt I was missing.
It worked. That is why this series exists.
This series is not about which model you should subscribe to or which Copilot tier is best value. The answer to that question changes every quarter, and someone else will write it better. This series is about mindset. About how you decide where AI fits in your own work. About how you structure your tools, your prompts, your agents so the small wins compound. About admitting where you are and starting from there.
On admitting where you are
If you have been quietly feeling the same thing I felt a few months ago, you are not alone. The first thing I want to say to you is simple.
If you are at the very early stage of using AI in your daily work, there is no shame in admitting that. None.
It is the first step. And it is a real step. The moment you stop performing fluency you do not have and start describing your actual position, every choice that follows becomes easier. You can ask the questions you actually have. You can pick the first task to try, not the most impressive task to demo. You can be honest with your team about what you do and do not know.
I have sat in rooms where the loudest voice on AI was also the person doing the least with it. I have sat in rooms where the quietest voice was running the most thoughtful, deliberate adoption I had seen all month. The two are not connected. Where you actually are has nothing to do with how confident you sound.
So if you are at the start, you are at the start. Say it. Then take the next step.
Why adoption is the most important part
Concept is interesting. Tech is exciting. Neither one changes anything by itself.
Adoption is what changes something. It is the day-to-day practice of pointing an agent at a real task, watching what comes back, learning what to ask better next time, and slowly, deliberately reshaping how you work.
A team that has read every AI white paper but uses none of these tools is in the same place it was last year. A team that uses the tools every day on real work has moved, even if its internal vocabulary is unfashionable. The work is the proof.
The reason adoption is undersold in the public conversation is partly that it is unglamorous. Nobody puts "I drafted a better project estimate with an agent today" on a keynote slide. But that is exactly the kind of small, repeatable shift that compounds into something significant over twelve months. Concept does not compound. Practice does.
What has changed for me
Over the last few months, I have been moving deliberately from talking about AI to adopting it.
In my own daily work, that has meant being honest with myself about where AI fits and where it does not. I started with tasks I would not have wanted on my plate anyway. Triaging incoming support tickets and drafting first-response replies. Turning a long discovery call into a structured requirements document I could then review. Generating first-pass effort estimates and budgets from a scope document. Cleaning up messy spreadsheets and surfacing the anomalies hiding in them. Designing a solution end to end and breaking it into clear implementation steps.
Some of these worked the first time. Some took three or four attempts and a lot of correction. A few I stopped doing, because the agent was reliably worse than just doing the work myself.
The wins are concrete. The misses taught me more than the wins.
What has changed at Equerra
The shift at Equerra has been the same idea, scaled.
We moved from talking about AI to using it on our own work first, before recommending any of it to a customer. Be customer zero. If you have not done it on your own delivery, you do not get to recommend it to someone else's.
What that looks like at our scale now is months of work on agents, prompts, skills, and the small automations that sit inside how we operate. Most of what happens here every day touches AI in some way. Not as a feature we sell. As a piece of how the work actually gets done.
The point is not novelty. It is reach. A small team using AI deliberately can now deliver the kind of output that used to need a team of thirty or more. That is the part that has surprised me the most.
Some concrete examples of what AI has taken off our plates. We do not answer the same question five times for five different customers any more. The agent trained on our own context handles the consistent ones. We do not manually trace a customer request across multiple systems to work out what is going on. The agent assembles the picture and surfaces what needs a human eye. We do not start a project estimate or a change-request scope from a blank page. The agent gives us a structured first pass, drawn from the way we have estimated similar work before, and we shape it from there. Even simple tasks like formatting a document to our brand standards happen without anyone touching them.
This is not arrival. We are not finished. The tech changes weekly, sometimes daily, and what was the right pattern three months ago is often not the right pattern today. We rebuild agents when a model upgrade changes how they behave. We retire prompts that no longer fit. We try ideas that do not work and quietly drop them. The discipline is in keeping the loop going.
I will write more specifically about what we have built and what we are still figuring out in later posts in this series. Some posts will be about how we do things at Equerra. Some will be from my own MVP journey, where the work looks different but the discipline is the same.
Why I am writing this series
I see two things at the same time right now. Both are real.
One. AI is moving fast, and the gap between people who have started adopting it and people who have not is widening, week by week.
Two. Most of the published content on AI is still about the technology, not the practice. There is plenty for someone who wants to read about it. There is much less for someone who wants to start doing it.
I want to add to the second pile. Not because I have all the answers. I do not. The journey is genuinely useful to share while it is happening, not after, sanitised, in a case study.
I plan to publish a series of posts over the coming months. Some will be reflective, like this one. Some will be specific and tactical, with the exact patterns I am using on a particular kind of work. Some will be honest about what is not working yet.
If you are starting your own adoption journey, I hope these posts give you something to compare against. If you are further along, I would love to learn from your patterns too. Drop a comment, send me a note, write your own post. The conversation gets richer when more people share what they actually do, not just what they think.
Where this series begins
This is the first post. The starting line.
If you are reading this and you are at the early stage of AI adoption, you are in good company. Start where you are. Pick one task. Try it. See what comes back. Try a better prompt next time. Keep the win, log the miss, move to the next task.
Adoption is a path, not an event. The people who started six months ago did not have a head start. They had a starting point. That is the only difference.
You can have one too. Today.
