AI in the application process: The whole world is talking about AI – but what does that have to do with my application?
„"Machine learning, generative AI, ATS, chatbots, agents, MCP interfaces" – heard of them? Maybe, but do you understand them?
ChatGPT, Gemini, Claude, Copilot and so on might be more like it – isn't that the AI I use to find recipes, conduct internet research, or even talk to in order to solve problems…?
My head was spinning at first when I tried to understand how this AI, this artificial intelligence, is used in the job market and what that has to do with my job applications! But now I understand it and I want to explain it to you as simply as possible.
Creating CVs and cover letters or practicing job interviews
Let's start with generative AI.
„"Generare" is a Latin word meaning "to create." Generative AI is therefore an artificial intelligence that creates or produces something for me. Examples include ChatGPT, Gemini, Claude, and many others.
With their help, I can, for example, create application documents such as CVs and cover letters, check my documents for spelling errors, or even practice job interviews using the voice function of these programs. Very practical!
The catch, however, is that these programs are so favorably designed that you almost always get positive feedback – even if the documents themselves aren't actually that good. Or, and this is even more likely, the documents end up being written far too perfectly, so that a person immediately recognizes: no human wrote this, or rather, this has nothing to do with my personal writing style.
Applications of generative AI also include chatbots, which allow us to contact an employer, for example, by simply entering a short message on the website and communicating with the AI. Companies like Deutsche Bahn already offer a job search via chatbot. Users navigate the system using largely pre-defined language modules to choose from. But woe betide anyone whose request doesn't exactly match a button that pops up! It's time-consuming and still has room for improvement, I'd say – I've tried it. The classic route via the company's career page is still faster. FOR NOW, at least!
Analyze resumes, identify key terms, and compare applications.
In contrast, with machine learning, the AI learns on its own based on patterns. How exactly does that work? The AI is first fed data – for example: round and red = apple, long and yellow = banana. Now the computer compares new images with this data and notices: "Oh, that's round, but it's green! Bananas are long and yellow, apples are round and red. That's not a perfect match... But because it's round, I'm 80 percent sure it's an apple." Then it gets the answer: "Yes, that's right, there are also green apples!" Now the AI has learned that green apples exist – and so it continues.
And what does that have to do with my application? Artificial intelligence, which learns through machine learning, is used by companies, for example, to analyze resumes, recognize keywords, and compare applications with their own job postings.
We can think of these so-called application management systems as a digital postman who collects all applications, quickly reads them, looks for important keywords, and then pre-sorts or even rejects them. Ultimately, the human reviewer reads primarily those applications that are the best fit.
This is very convenient for employers – but still exists in a legal gray area. AI can certainly discriminate if it doesn't know exactly what to do. And some good candidates might simply slip through the cracks. But it's probably only a matter of time before these applicant management systems become the rule rather than the exception.
Use related job titles and derive skills
This machine learning AI includes, for example, ATS systems (Applicant Tracking Systems), which, as described above, are now being used by an increasing number of companies. This also includes the matching systems on job portals.
A job board like Stepstone can therefore understand, for example, that when searching for a "project manager" job, job postings for "project lead" or "product owner" might also be relevant. The AI thus understands related job titles.
Indeed offers an instant match feature with its resume database, allowing employers and candidates to connect more quickly. This feature even goes so far as to infer additional skills from the entered data, similar to LinkedIn. For example, if I tell the program that I'm skilled in content marketing, the system automatically deduces that I might also excel in copywriting or SEO.
So AI really does help me get more out of my search! Occasionally, though, it overshoots the mark, and when searching for a "job coach" position, "special education teacher" also appears in the matching results. Oops! But of course, this mainly happens when the relevant job board isn't currently offering much of what I'm looking for.
Agents automatically search for suitable job offers.
Finally, a word about agents and MCP interfaces… No, not 007, but still not bad!
In the future, the near future… no, it's already happening in reality – I tried it out just recently during a training course! We communicate A:A, meaning agent with agent. So the AI talks to the AI before any human even takes over.
Contracts are negotiated machine-to-machine until the AI withdraws at a time previously set by humans – hooray! – and lets us take the reins again. Unsolicited applications might soon work this way. But it's still a bit of a pipe dream – although AI is so fast that it could happen sooner than we think.
Fully automated job search with agents and MCP interfaces
We can already make excellent use of agents. You'll find them next to the text field in the AI. For example, I can assign an agent to send me job offers every day that perfectly match my profile. So, it searches for me. That's real added value – fully automated!
We will likely soon benefit from the so-called MCP interfaces. MCP stands for Model Context Protocol and is also referred to as the "USB-C port for AI." We can think of it as a cable that connects our chosen AI to all the apps we already use.
Then we simply choose which AI to use – Gemini, Copilot, ChatGPT, Claude…? It's then connected to our other programs: our email program, social media accounts, LinkedIn, Xing, and other applications. I can then interact with the AI like an employee and train it as an agent that handles my bookkeeping or automatically sends out my applications as soon as a suitable position becomes available.
My conclusion
AI is currently a tool that allows me to move things forward very quickly. It's so fast that it makes my head spin again. It can be incredibly fun, and I don't always have to try so hard.
But let's be honest: it does its work based on the data it's been fed. While this data is extensive, it's not necessarily complete or accurate. AI still communicates in ways that don't always resemble human speech. It can certainly discriminate and even hallucinate, essentially making things up.
In the context of the job application world, we should definitely use them to improve the matching with job postings, to get past the digital mailman and ultimately to reach the person I want to work for – or rather, to find the person who fits my company within the large crowd.
AI conclusion
I asked Copilot:
„"How long do you think it will take until AI has automated the application process to such an extent that a person only needs to show up for the interview – provided they have submitted their CV?"“
The answer on October 5, 2026:
„"If you ask me when a typical application process looks like this, where someone simply uploads their resume and only sees another person for the interview, I would say:
- In large companies: In some cases, already today.
- Over the next three to five years: probably in many standardized professions.
- Over the next five to ten years: This could become the norm in a large proportion of office, sales, service and administrative professions.
The real bottleneck isn't the technology. AI can already analyze resumes, understand job postings, rank candidates, create competency profiles, generate suitable interview questions, evaluate aptitude tests, and analyze video and audio interviews. All of this already exists.
What still holds companies back today are data protection, the EU AI Act, discrimination risks, liability issues, acceptance among applicants, works councils, and the shortage of skilled workers – sometimes they prefer to do the checks themselves. That's why I don't believe that people will disappear completely from the process.“
Katrin Bätz, October 5, 2026
