There seems to be an AI for everything now.
An AI to write the email.
Another one to improve the email.
One to find the prospect.
One to research the prospect.
One to score the prospect.
One to take notes during the meeting with the prospect.
One to summarise those notes.
And probably another one to tell you which of the other AI tools you should have been using in the first place.
I use AI myself.
ChatGPT, Claude and Gemini have become useful parts of how I research, prepare outreach, develop ideas, improve workflows and work through business problems.
So this is not an argument against AI.
Quite the opposite.
I think these tools are extraordinary.
But lately I have been thinking about something else:
At what point does AI productivity become AI fatigue?
And if almost everyone eventually has access to similar technology, what is actually left for us to compete on?
We've moved from AI curiosity to AI abundance
A few years ago, being able to use generative AI effectively felt like an advantage.
Today, it increasingly feels like basic professional literacy.
Marketing teams are generating content with AI.
Sales teams are researching prospects with AI.
Founders are building prototypes with AI.
Customer-service teams are deploying agents.
People who previously needed specialist software knowledge can now describe what they want and have a tool help them build it.
That is a remarkable shift.
But abundance creates its own problem.
When everyone suddenly gains the ability to produce more, more stops being particularly valuable.
More emails.
More posts.
More landing pages.
More videos.
More outreach.
More reports.
More presentations.
More personalised messages that somehow still feel strangely impersonal.
We solved the problem of production speed.
We haven't necessarily solved the problem of being worth someone's attention.
Marketing has a quantity problem disguised as a technology solution
AI has made content creation dramatically easier.
That should theoretically mean better marketing.
Yet look around.
Our inboxes are not suddenly filled with brilliant emails.
LinkedIn is not universally more insightful.
Every company has not magically developed a distinctive point of view.
In fact, sometimes it feels like the opposite has happened.
We have become extremely efficient at producing competent-looking sameness.
Recent industry research captures this contradiction quite well.
AI adoption among marketers is now widespread, yet Salesforce's 2026 State of Marketing research found that a large majority still reported running generic campaigns.
That fascinates me.
We have increasingly sophisticated technology capable of personalisation, research, automation and prediction.
And somehow we can still end up saying:
Hi [First Name], I noticed your impressive background...
at industrial scale.
The technology got smarter.
The message didn't necessarily get better.
Maybe the bottleneck was never writing the sentence
For years, businesses assumed that execution was the hard part.
If only we could produce more content.
If only the sales team could research every lead.
If only somebody had time to write all those emails.
AI helps solve many of those constraints.
But remove the execution bottleneck and another one becomes obvious:
Do we actually know what is worth saying?
That is a very different problem.
AI can give me twenty versions of a campaign headline in seconds.
It cannot decide, without context and direction, what this organisation should stand for.
It can help analyse a market.
It cannot take responsibility for deciding which market the company should enter.
It can help draft a sales message.
It cannot fully understand the history behind a difficult client relationship unless somebody provides that context.
It can suggest an objection response.
It cannot look across the table and sense that pushing harder right now will lose the deal.
These are not simply content-generation tasks.
They require judgement.
Sales makes the difference particularly obvious
Imagine two salespeople approaching the same potential customer.
Both have access to AI.
Both can generate account research.
Both can identify the likely decision-maker.
Both can create an outreach sequence.
Both can ask AI for potential objections and recommended responses.
So who wins?
Probably not automatically the person with the better prompt.
Maybe it is the person who listens properly.
The person who realises the problem described in the first meeting is not actually the real problem.
The person who knows when not to sell.
The person who remembers something important the customer said three conversations ago.
The person who can challenge an assumption without making the client defensive.
The person who admits:
I don't think we're the right solution for this.
That last sentence might lose a transaction.
It can also build trust.
And trust is difficult to automate.
Customers don't only buy information
This is something I think we sometimes forget when discussing AI.
A customer can increasingly research almost everything before speaking to a salesperson.
They can compare products.
Read reviews.
Ask an AI assistant for recommendations.
Analyse pricing.
Generate questions for the sales call.
Understand terminology that previously required talking to an expert.
That changes the role of sales.
If the salesperson's only value was giving the customer information, then yes, that role has a serious problem.
But good sales has never really been about transferring information.
It is about helping someone make a decision under uncertainty.
That involves:
context,
confidence,
negotiation,
risk,
trust,
and sometimes emotion.
The buyer is not only asking:
What does your product do?
They are also asking:
Do you understand what I am trying to achieve?
Can I trust your recommendation?
What happens if this doesn't work?
Will you still answer the phone after I sign the contract?
Those are human questions even when AI is involved in answering them.
The uncomfortable part: some work will disappear
I also don't think we should comfort ourselves with the easy line that:
AI won't replace people. People using AI will replace people who don't.
Maybe sometimes.
But it is too neat.
Some tasks will disappear.
Some roles will become smaller.
Some teams will need fewer people to produce the same output.
Some skills that once commanded a premium will become commodities.
If producing a competent first draft takes thirty seconds instead of three hours, the economic value of producing a competent first draft probably changes.
Pretending otherwise does not help anyone prepare.
The more useful question is:
Which parts of my work become more valuable when execution becomes cheaper?
That is where I think the conversation gets interesting.
What becomes scarce when production becomes abundant?
I keep coming back to a short list.
Judgement
Knowing which recommendation to follow, and which one to ignore.
AI can generate options.
Someone still has to decide.
Context
Understanding why something that worked in Australia might not work in Spain.
Why the same sales approach can succeed with one stakeholder and fail completely with another.
Why technically correct advice can still be wrong for this particular situation.
Taste
AI can produce endless variations.
Taste is knowing when to stop.
What belongs.
What doesn't.
What sounds like everyone else.
What actually feels like your organisation.
AI can generate options. Someone still has to decide.
Curiosity
The quality of an answer often depends on the quality of the question.
People who stay curious will probably extract much more value from AI than people who simply ask it to complete tasks.
Empathy
Not performative empathy generated at the end of an email.
Actual empathy.
Understanding why somebody is hesitant.
Why a team is resistant.
Why a customer is frustrated.
Why a stakeholder needs to feel heard before they are ready to discuss a solution.
Trust
Sales and partnerships ultimately involve people taking risks with other people.
AI can support that relationship.
It cannot simply declare itself trusted.
Trust has to accumulate.
Accountability
Perhaps the most important one.
AI can recommend.
AI can generate.
AI can analyse.
But when an important decision goes wrong, somebody still has to say:
That was my decision.
Responsibility has value.
Perhaps AI should make us more human, not less
The optimistic version of this future is not one where humans try desperately to compete with machines at machine-like work.
That sounds exhausting.
AI is already better than me at many things.
It can read faster.
Calculate faster.
Summarise faster.
Generate alternatives faster.
Search through information faster.
Good.
Let it.
The opportunity is to use that saved time for the things we regularly claim are important but often neglect because we are too busy producing things.
Talking to customers.
Building relationships.
Thinking properly.
Testing assumptions.
Coaching people.
Understanding the market.
Improving the product.
Being present in a difficult conversation.
Doing the work that requires us to actually know another human being.
There is a strange possibility here.
Technology that appears to make work less human could force us to become better at the parts of work that are uniquely human.
The advantage won't be having AI
Soon, saying:
We use AI
may sound a little like saying:
We use email.
Of course you do.
The interesting questions will be:
What do you use it for?
What do you refuse to delegate to it?
How do you know when it is wrong?
What do your people do with the time it saves?
Has it improved the customer's experience?
Has it improved your decisions?
Has it made your work more distinctive?
Or have you simply become faster at producing more of the same?
That is the part of AI adoption I find most interesting.
The technology will keep changing.
There will always be another model.
Another agent.
Another platform.
Another tool promising to automate the job that the previous tool had only just transformed.
Trying to keep up with every one of them sounds like a very efficient route to exhaustion.
So I am increasingly less interested in asking:
What's the newest AI tool I should be using?
And more interested in asking:
What is the part of this job that I should still be getting better at?
Because there will probably always be an AI for the job.
The more important question is what we choose to bring to it.