Technically possible, but conceptually questionable

“Software development at t2informatik” – that was the subject line of an email we received last week.

What made this application stand out was that the applicant himself pointed out that it had been generated entirely by AI, based on the content of our website, and then sent automatically. At the same time, he saw the application as a practical example of his skills: a Java/Spring-based outreach workflow that identifies suitable companies, analyses content and automatically generates applications from it.

Technically, it was certainly interesting.

And yet, I was left feeling less admiration than irritation.

As much as the application was intended to demonstrate that the applicant understands AI and modern software development, it revealed little else: a genuine interest in us. In our work and our culture. In the question of why t2informatik, in particular, might be a good place for him and why he would be a good fit for us.

For me, this is precisely where a key tension lies in dealing with AI: today, with surprisingly little effort, we can produce things that are formally correct, linguistically sound and technically impressive. But not everything that can be generated is therefore meaningful, helpful or convincing.

Some things are technically possible, but conceptually questionable.

Some examples and AI as a means to an end

An automated job application is one example of new ways to use AI for specific tasks. But it is by no means the only one.

• Do you want to network with people on LinkedIn and lay the foundations for a business relationship? No problem.
• Would you like to cold-call people using a voice-activated computer? Technically possible.
• Do you want to interact with people who visit your website? The chatbot is available round the clock.
• Would you like to sell your products by using information scraped from websites for cold-mailing campaigns? It’s a piece of cake.

In these and many other examples, AI has long been a means to an end. It consolidates information, writes texts and executes workflows. That can be impressive. It can save time. It can streamline processes. What a wonderful new world.

But wait: What problem does AI actually solve here?

Obviously, it initially solves a problem of volume and time. On LinkedIn, there are a great many people who could, in principle, be of interest to the company’s services. Contacting them all manually takes far too long. Out in the wider world, there are even more people with a telephone line. As sales agencies are relatively expensive, using AI is naturally the obvious choice. Customer support only operates during standard business hours, but the chatbot knows no working hours, no holidays and no bathroom breaks. And what about email campaigns? If there’s one thing large language models are good at, it’s generating text.

Seen in this light, AI sounds like a fairly obvious solution. A wonderful new world.

But does the end really justify the means? And what problems might AI actually fail to solve after all?

What is the real purpose?

If AI is the means, what, then, is the real purpose?

• Connecting with large numbers of people on LinkedIn?
• Reaching out to potential prospects by telephone?
• Providing expert answers to unresolved questions?
• Or sparking interest among potential customers?

That is not the purpose. We wouldn’t need AI for these things. We could just as easily contact people manually, ring them up, answer their questions face to face, or personally draw their attention to our products. So the action itself is not the reason why we do it.

The real purpose lies one level deeper: relationship.

We don’t network to fill a contact list, but because we hope that something sustainable will eventually come of it. We don’t make calls to work through a script, but because we want to find out whether there’s a basis for working together. We don’t answer questions to close deals, but because we want to show that someone cares. We don’t spark interest as an end in itself, but because the ultimate goal is a collaboration that works for both sides.

And this is precisely where the real problem lies: AI can deliver the action, but not what that action is meant to be built upon. It can connect people without fostering a relationship. It can make calls without building trust. It can reply without making anyone feel understood.

This is not a technical problem. It is a confusion of means and ends.

Anyone who loses sight of the end goal will eventually come to regard the frequency of the action itself as success. More LinkedIn contacts, more phone calls, more tickets resolved or more emails. The numbers go up, and yet not a single solid relationship may be formed.

But AI can already do so much

But, but AI can already do so much. And in future, it will certainly be able to do much more. And do it better than most people can, too.

That’s true. AI is already achieving a great deal and supporting people in all sorts of work-related tasks. Anyone who struggles to write short and concise emails is delighted by the truly miraculous sentences produced by these nimble text generators. Anyone who can’t be bothered to write page-long summaries of meetings is desperately hoping for technical support to conjure up digital paper monsters as if by magic. And anyone who doesn’t know what a conversation guide for an initial meeting might look like simply types in their problem, only to try out the guide straight away during a meeting with a prospective client.

All of this can be helpful.

AI isn’t the problem. Nor is automation the problem. It becomes problematic when we act as though the technical generation of an action can already replace what that action is actually meant to achieve.

Relationship.

That is the magic word that it’s really all about. Relationship is the foundation for the purpose of the examples I’ve repeatedly mentioned. Without a relationship, everything that comes from AI or is created with the help of AI is of little value. It is simply output. Quickly generated, easily documented. Yet without any particular value and, with very few exceptions, without any real impact.

Shall we take a slightly closer look at this?

Relationships as the foundation for everything positive that follows

‘People buy from people’ is not exactly a new insight. Whilst it may not hold true in 100 per cent of cases, it does hold true surprisingly often. When is it true? When there is a relationship between the parties involved. When there is a certain degree of trust. When the relationship is built in an appropriate manner – for example, competently, in a friendly and human way.

What do many people do after they’ve automatically connected with others, without them even realising it? They send automated messages using AI. Formally correct, friendly in tone, technically flawless. With a seemingly personal touch. And yet often without any genuine interest.

What’s more, many people now realise quite quickly whether they’re being written to by a human or a machine. They recognise the patterns, the polished phrasing and the apparent personalisation. And as soon as this impression takes hold, the opposite of what was actually intended happens. Instead of trust, distance sets in.

Can a relationship be built this way?

My clear answer: No, that’s not how it works.

A relationship is something that develops. It’s not a quick fix. It’s not a one-hit wonder.

Could AI be of help here? Of course. But not as a means of handling large volumes or saving time, rather as targeted support. In formulating that one question you want to ask a specific person because you’re interested in the answer to that question. This question can form the basis for an exchange between two people. It can lay the foundations for a relationship that might eventually lead to a business relationship.

Conclusion: Technical feasibility is not enough

The automatically generated job application was technically interesting. It demonstrated what is possible today with AI, website analysis and workflow automation. And that is precisely why it was a good example of the tension we are currently facing.

AI can help. It can summarise information, draft texts, organise thoughts, provide conversation starters and support people in phrasing things better, structuring their thoughts more clearly or communicating in a more targeted way. It can also be useful in building relationships: when preparing for a conversation, researching a company, formulating a good question or identifying a potential common ground.

But it cannot replace the actual process of building relationships.

A message does not become convincing simply because it sounds good on the surface. A phone call does not become valuable simply because it is initiated efficiently. A job application does not become suitable simply because it was generated on the basis of a website. What matters is whether genuine interest is evident behind the action – whether a person is prepared to really engage with another person, a company, a situation or a matter.

This is precisely where the line is drawn between meaningful support and dubious automation.

Anyone who uses AI to listen better, look more closely, ask smarter questions or formulate thoughts more carefully can improve communication. Anyone who uses AI merely to simulate interest, closeness or a relationship produces one thing above all else: output. Perhaps a lot of output. Perhaps even impressive output. But not necessarily impact.

Ultimately, therefore, the crucial question is not: Is it technically possible?

The crucial question is: Does it make sense in terms of content?

Just because something is technically possible doesn’t mean it isn’t questionable in terms of content. It’s up to you to do better.

Notes:

Michael Schenkel has launched a “Blog parade: AI and my job”. You can still take part until 24 July 2026 at 12:00.

Would you like to support Michael Schenkel or discuss the sensible use of AI beyond bulk operations? Then share this post on your social media channels.

Michael Schenkel has published more posts on the t2informatik Blog, including: