Do you need help & advice with AI & Automation?
We have written about this thing three times and taken a different view each time. That is a fair reflection of how most people feel about it, and it is not much use to anybody, so here is the honest version.
Nobody knows. We do not know what our own work looks like in three years, and we are the ones who are supposed to be advising on it.
Will AI take my job?
Some jobs, yes. Probably not in the way or the order anyone is predicting.
This is not new and it is not a reason to panic. When the motor car replaced the horse, it took a few years, and then a whole set of jobs simply did not exist any more. Nobody who worked with horses saw a memo about it.
More recently: video shops. There used to be one on every high street with people working in it. Nobody sat down and decided to end that.
What is different this time is the speed and that it is reaching work people assumed was safe. If you are a lawyer, an agent may be able to do a meaningful part of your job sooner than feels comfortable. That is a genuinely unnerving thought and pretending otherwise is not honest.
Our honest position. We are nervous and excited in roughly equal measure. Some of what we do today will be done by software. We do not know which parts, and neither does anyone else, whatever confidence they say it with.
So should I be worried?
Worried is not useful. Curious is.
The people who came out of previous shifts badly were not the ones whose jobs changed. Everybody’s job changed. They were the ones who waited to be told what it meant.
The practical difference is small and it compounds. Somebody who spends an hour a week working out what these tools can actually do ends up, within a year, being the person others ask. That is a considerably better position than being the person waiting for a decision.
What actually works, in practice?
Collaborating with it. Not handing work to it.
This is the most useful thing we have learned and it is the opposite of how most people start.
Left to get on with something by itself, AI makes too many mistakes and gets stuck. Given a specific job, with you checking as it goes, it is genuinely good. The difference is not the tool, it is whether a person stays in the conversation.
| Works | Does not |
|---|---|
| Give it a defined task and check as it goes | Hand over something vague and hope |
| Ask it to draft, then you edit | Ask it to finish, then you send |
| Use it on work you could have done yourself | Use it on work you cannot check |
| Treat it as a capable junior | Treat it as an expert |
That last line is the useful frame. You would not let a junior member of staff go and spend thousands of pounds on flights without a process around it. Not because they are stupid, but because nobody new should be making unsupervised decisions that cost money. Same principle, and it is not an insult to the technology.
What does this mean if I run the business?
Two things, and neither is about headcount.
First, the businesses getting the most out of this are not spending less on their people. They are getting more from them. The work that goes is the work nobody enjoyed: the drudgery, the retyping, the chasing. What is left needs more judgment, not less.
Second, your staff are already thinking about this, and most of them have not said so. Some are quietly worried and some are quietly experimenting. Both are more useful to you out in the open. A business where people can say “I used AI for this bit” is one where you can see what is happening. A business where that feels risky to admit is one where it happens anyway, invisibly.
[VIDEO_PLACEHOLDER]
So, what now?
Whatever your job is, put some time into understanding how these tools actually work. Not the headlines, the tools. An hour a week is plenty and it compounds faster than you would expect.
If you run the business, say out loud that using AI is fine and being open about it is expected. That one sentence changes what you can see.
And if you want somebody keeping track of where this is going on your behalf, that is what an AI Manager is for. The practical starting points are in getting started with AI in your business.
What do you think your job looks like in three years? We would genuinely like to know, because we have no idea about ours.
Frequently asked questions
Will AI replace jobs?
Some, yes, as every significant technology has. Which ones and how fast is genuinely unknown.
Be sceptical of anyone confident in either direction. The people saying nothing will change and the people saying everything will are both guessing.
Should I tell my staff AI will not replace them?
Only if it is true, and you cannot know that.
A promise you cannot keep costs more than an honest conversation. What you can say truthfully is that you are not planning to reduce headcount because of it, if that is the case, and that being open about using AI is expected rather than risky.
Which jobs are most exposed?
Work that is mostly producing routine text or moving information between systems.
Though “exposed” usually means parts of a job rather than the whole of it. Most roles are a bundle of tasks, and AI is taking tasks, not roles, at least so far.
What should I do to stay useful?
Learn how the tools actually work, and get good at judging output rather than producing it.
The scarce skill is turning out to be knowing when something is wrong. That is not a technical skill and it comes from knowing your own job well.
Are AI agents going to do all this on their own?
Not reliably, not yet, and the ones that try tend to get stuck.
Agents are genuinely useful for narrow, well-defined jobs with a person checking. Open-ended autonomy is not where the technology is.
Last reviewed: 17 August 2026.
Related
- Who actually owns AI risk in your business
The silence at the management meeting, and who the job should go to.
