As both an active participant and an observer of the AI industry I have been closely watching this weird phenomenon unfold over the past couple years, and I think it’s kind of hilarious.
Do you remember the good ol’ days when job interviews used to go like “Sir, do you know React?”. . . but nowadays they are more “Cool, but can you make presentations, market research, and how about your process analysis skills, and when you’re at it, could you have our AI chatbot stop yelling at customers?”
The whole professional landscape has done this bizarre flip where being really, really good at one specific thing now all of a sudden feels kinda quaint – I mean, knowing all the words to “Ice Ice Baby” kinda quaint. Companies don’t want specialists no more because they want these mythical creatures that are called ‘AI generalists’ who are able to MacGyver* their way through any problem with nothing but an AI and an API key.
And I am totally here for it.
* McGyver, 1985 TV-series about a dude who solves world ending problems with duct tape and chewing gum.
More rants after the messages:
Join me and Eliska Went on September 5th when we show you how to create your own automated AI tool to find, write and apply for jobs. Register here. Here, I mean: here
- Comment, or share the article; that will really help spread the word 🙌
- Connect with me on Linkedin 🙏
- Subscribe to TechTonic Shifts to get your daily dose of tech 📰
- Visit TechTonic Shifts blog, full of slop, I know you will like !
What the heck is an AI Generalist?
Here’s where most people get it wrong. An AI generalist of course is someone who knows a bunch of AI tools, but they are also the people who use AI to move above and beyond their original job description and become something like a modern homo universalis*
Think about it.
You started as a marketing manager, segmenting customers and preparing data for the company’s daily email blasts. This work is not going to get you are raise anymore, because you now have tools like Adobe taking over the entire workflow. But you were a smart-ass, because now you are using AI to conduct deep market research with tools like SciSpace for scientific papers and Exa Websets for basically everything else. You are building process automation workflows that make the IT department freaking jealous, and you’re creating app prototypes in Lovable, a vibe coding app, you just automated your segmentation job with Zapier or n8n, so now you have the time to design presentations in Manus or Prezi to show off the ‘value add’ that you are! And when the IT department acts up, trying to undermine your ass, you just fire up Google’s screen-sharing AI (Google Stream) instead of filing a help desk ticket like the rest of the peasants out there.
But this is not all. . .
Because before every meeting, you create AI-generated briefings in a beautiful format (I use Manus again) so you show up prepared instead of winging it for an hour and everyone is bored to death and pretend to pay attention. Basically you are not only doing the job you were supposed to be doing, but you are doing your job plus understanding the context around it, plus fixing the processes that make everyone’s life harder, plus building tools that didn’t exist yesterday.
I watched this play out recently when I was teaching a bunch of engineers about the possibilities of AI. The people who excelled in my class weren’t the machine learning wizards nor the people with PhD dissertations, but they were the curious weirdos who kept poking at the AI until it started to make sense.
Now that is the secret sauce right there people, professional-grade curiosity mixed with the audacity to think you can do everyone else’s job too.
* The Renaissance ideal of knowing enough about everything to be dangerous in the best possible way.
Why organizations suddenly want these people
Let me paint you a picture of what I see is happening out there.
Companies who are active in software development are sitting on legacy code that’s older than some of their employees, but nobody wants to touch the ancient .NET applications (or beyond), but they are not going anywhere either. Enter the AI generalist, who uses Kiro* to debug old Java code he has never seen before and generate unit tests while he’s looking at his Reddit feeds. But he is not stopping there and he’s also analyzing the business processes that created this awful mess in the first place and he proposes automation solutions to prevent future disasters.
I know a guy that works this way. His name is Thomas Mann. He uses AI to manage the context of his business and to boldly go where no other dev has gone before. He is currently engaged in a consulting gig to teach devs how to be like him. Oh, because of AI he now has time to air in our podcast. Click link.
Healthcare is getting absolutely weird too. Hospitals want people who can wrangle AI diagnostic tools without accidentally prescribing meditation for appendicitis. But the smart ones are also using AI to map patient flow processes, and they are identify friction points at the task level, and build prototypes for better systems. They are conducting their own research instead of waiting for some consultant to tell them what’s broken. A good example is Tom van der Laan, he is an ENT doctor at UMCG and he recently got caught quoting “It’s wonderful to see what artificial intelligence is capable of”, when he demonstrated the AI based translation and transcription systems of Timekettle. Van der Laan continues, “If I have clinical work for half a day now, it takes me about 1.5 to 2 hours of preparation time. I read the file of each patient, examine changes in medication use and treatments by other healthcare providers. If AI can summarize that with only the relevant information, it can save a lot of time
Banks are even funnier. They have had fraud detection systems for years, and now they have added compliance bots, and customer service chatbots that need constant babysitting, and here the AI generalists are maintaining these systems and they are creating portfolio management tools that rank projects by risk and value, they’re building cost analysis models, and they’re generating architecture recommendations for process automation. So one day you were working with spreadsheets, and the next you are presenting a video explanation of your strategic recommendations using HeyGen because you want everyone to get the same message.
Schools are probably having the most fun with this. Teachers are becoming seisoned AI prompt designers and they’re conducting deep research into learning methodologies, and to keep up with the latest developments in their field, they’re building educational apps, and they’re creating automated assessment systems. They have become a research department, IT support, and curriculum development team all rolled into one person. I am a part time university teacher and I am teaching my students how to add value as a human to use the AI to your advantage and not to be discouraged by it. My entire curriculum slide deck has been generated by AI (using Manus and Genspark).
And now, even creative industries are going full homo universalis. They tried to withstand the Hunns at the gates, but now they are finally embracing the AI as a partner in their business. They have to, because more and more of their customers are using AI themselves from generating webtexts, to AI vids, and automating campaigns. You now have a new category of generalists called iWriters, and they are using AI to brainstorm, and also to analyze market trends, building competitor research databases, creating business process maps for content workflows, and prototyping interactive experiences. The smart ones treat it like becoming a one-person creative agency.
* Kiro was Amazon’s internal app that saved them thousands of hours of dev time for their upgrade from old Java 7 to the latest release, and now it is commercially available (though a waiting list)
Why this is happening right now
When I started my career, the path was beautifully simple.
Pick a career.
Master the craft in 10.000 hours.
And become the go-to person for that one specific thing. It worked great for me for about twenty years, until AI showed up and started erasing the boundaries between departments like a tactical nuke.
Now I watch developers generate database schemas with AI assistance. Data analysts are asking language models to write their SQL queries. Content managers are connecting AI summarizers to customer relationship systems, then building process automation workflows that eliminate half their manual work. Everyone’s borrowing from everyone else’s toolkit, and the old departmental walls look about as sturdy as a house of cards in a hurricane.
The barrier to entry has dropped so low it’s practically underground. You can test the latest AI models with nothing more than a credit card and a flimsy internet connection. Some are even free, which still blows my mind though, and companies don’t need research labs anymore. They need people who can tinker without breaking everything and build solutions faster than procurement can approve software purchases or at least faster than IT can ban them for being shadow AI.
You know the “bring your own device” mantra from back in the day? When you were suddenly allowed to bring your Apple into the workplace. Well, now it’s going to become “bring your own AI”.
The forward deployed engineer
I thought the job market couldn’t get any weirder, but then I read about the Forward Deployed Engineer.
Dawhudnow?
Well, that is a role which is so confusingly multifaceted, and it is making specialists in IT obsolete in a number of companies already. They are the people who somehow convinced companies they needed software engineers who also do sales, customer success, and consulting, the cool thing is that they do all of this while still maintaining the ability to code without having a nervous breakdown.
Say you are a software engineer, but you don’t want to hide yourself in your cubicle with noise-canceling headphones. You are instead flying to Iowa to sit on a tractor with farmers, and there you’re trying to figure out how to make John Deere’s AI stop recommending pesticide for corn that doesn’t exist.
One week you’re debugging code at the headquarters of your company, and the next you’re embedded with a customer’s team in some industrial facility, where you are pretending you understand their 20-year-old legacy systems and you’re building prototypes that somehow need to integrate with everything.
The role was basically invented by Palantir, a.k.a. ‘the Beelzebub of AI’, which should tell you something about its inherent weirdness. They called their Forward Deployed Engineers “Deltas” (as in ‘Delta Force’ – awkward nerd alert!) and they now have more of them than they had regular software engineers for years. These people can essentially be compared to startup CTOs, who are continuously bouncing between customer requirements and product teams, and building bespoke solutions and improving the core platform at the same time. It can be compared to a consultant who actually has to make the code work long-term, which is either the best job ever or a special kind of professional torture.
I believe it is the latter.
You didn’t see that coming now did ya?
Lemme mansplain . . .
The true meaning of passion
The people who can pull off this feat – to use AI to augment them in such a way they are going above and beyond their role description – they are the passionate ones among us. Because not all of us will want to stick their heads above the parapet.
People, mostly CEO’s or HR execs, throw around the word ‘passion’ like it is confetti at a Tony Robbins or Stephen Covey seminar for managers, but the word literally means ‘suffering’*. It is not the Instagram-friendly kind of suffering where people post their late night work sessions, but actual soul-testing and character building endurance.
The people who can pull off this AI generalist feat I am talking about in this blog post, the ones who are transforming themselves into nouveau renaissance types where they are using AI to transcend their job descriptions, they are the passionate ones among us in the truest sense. They are the ones who are willing to suffer through the learning curve of yet another AI tool that promises to revolutionize everything but mostly just breaks in creative new ways. They endure the frustration of building process automation workflows that work perfectly in testing but somehow summon demons when deployed to production. That is real passion, of the original kind. The kind that means you are willing to undergo the discomfort of constantly learning, and to suffer through the inevitable failures that come with pushing boundaries, and they are definitely not the people who talk about passion on LinkedIn while posting stock photos of keyboards. These are the people who are actually sick with it, in the most etymologically accurate way possible.
* I just like etymology, because it reveals the true meaning of a word. ‘Passion’ comes from the Latin ‘passio’ which literally meant ‘suffering’ and ‘endurance’. Early Christians used ‘passio’ to depict Christ’s suffering (hence ‘the passion of the Christ’), and it’s cousin is the Greek ‘pathos’ which also means ‘suffering’, and ‘experience’, and from that root you also get ‘pathology’, ‘pathetic’, and ‘sympathy’ and ‘apathy’, etc. But in medieval Europe the word passion broadened. It went from ‘suffering’ into ‘strong feeling’ or ‘overpowering emotion’. I think because pain and desire both feel like you have been hit by a truck – but who am I – I ain’t no linguist. So, love, rage, lust, zeal . . . all could be ‘passions’.
The skills that actually matter
I have spent for over three years now crawling elbow-deep in this generative-AI cesspool, and I have figured out what actually makes someone useful in this new world.
First up is prompt engineering, which sounds fancy but is basically learning how to talk to the AI without confusing them. You need to be specific, give context, and make your requests repeatable. It’s not magic, and you certainly need not buy an eLearning course on Coursera or Udemy, or even worse – hire me as a consultant to help you write short text. Just think of it like this and you’re gonna be fine: the AI is your Oompa Loompa and you have no hands and a limited brain, but your Oompa does, and the only thing you can do to interact with the outside world is to tell it to do things for you as if it is your highly personalized, personal assistant – which it basically is. And that is how you talk to the AI.
But hey, if after three years you still are struggling with ‘prompting’ you might skip this blog post altogether cause nothing’s gonna save you from extinction.
Here is where things get funky. You also need to become comfortable with deep research tools that go way beyond Google or ChatGPT deep search. Try for instance SciSpace, that lets you dive into academic papers and extract insights without having to read 200 pages of methodology, and Exa with their Websets tool can map competitive landscapes in ways that would take traditional researchers weeks to accomplish.
Process analysis becomes your next superpower. You can use eDrawMax or just prompt Manus AI to map out business processes, based on your textual description of them. Of course ChatGPT will also try it if you ask it nicely, but it will then ask for Level 3 friction analysis – that granular task-level view where the real productivity gains hide, and in the end it will generate a bunch of code called Mermaid blocks, which you have to paste into mermaid. live to render a diagram. Cumbersome as hell, so the go to for me is either eDrawMax or Manus or Genspark. Those AI can even suggest automation architectures and cost estimates, though your business experience still determines what’s actually worth implementing.
Prototyping tools like Lovable let you build functional applications without having to wait for the developers to help you, and presentation tools like (again) Manus AI or Prezi’s AI tool help you communicate complex ideas visually based on your textual description of it, and HeyGen lets you create video explanations that deliver consistent messaging even when you’re not in the room.
Most importantly, you need to understand business context to maximize your value add to the company you work for.
A chatbot that saves customer support ten hours per week is worth more than the fanciest dashboard that nobody uses. But an AI generalist goes further because they understand why those ten hours matter, what else could be automated in that workflow, and how to measure the broader organizational impact.
How to become one of these people
Here’s my totally unscientific but surprisingly effective approach to joining this AI generalist club.
Start with a problem you actually care about (remember: caring = passion = suffering), but don’t stop at the obvious solution. Maybe you hate tracking expenses, but instead of just finding an expense app, use the AI to analyze your spending patterns, build custom automation workflows, and create predictive models for budget planning.
Yes, you yourself can create predictive models.
Just use Claude’s Sonnet from Anthropic because it is the most analytical general purpose LLM out there, and it generates the most beautiful visuals you can copy paste in your presentation, or if you want to go full nerd-mode, simply fire up Julius AI and have it work out the most complex analysis you can think off including error metrics, though I must admit their presentation skills are a bit below par. And if you don’t understand diddley squat about data science, just ask any AI. They’re more than happy to explain the results you’re seeing on your screen.
When you want to create a product or strategy plan, go full deeper than deep research mode. Use AI tools to conduct competitive analysis – market positioning, technology stacks, app types, anything that’s publicly available. Then turn your attention inward. Map your organization’s processes, identify improvement opportunities, and prototype solutions before proposing them. And meanwhile there are even more specialized research tools out there. Take for example FutureHouse’s AI platform, it is a public facing AI (-driven) research tool, you can use for molecular prediction or synthesis planning. Yes, the stuff you always wanted.
And document everything that works. Keep a personal collection of prompts, workflows, research methodologies, automation scripts, and of course the outcomes. Over time, this becomes your secret weapon, you know, a curated toolkit that reflects your unique approach to problem-solving. Mine are all stored in Notion, that is a note-taking platform quite similar to how the internet is organized. But there’s also free initiatives like Obsidian (even with AI integration) or GoogleLM – but beware.
If something is free, you’re the product. Or in this case, your research.
And above all, stay curious instead of trying to achieve perfection in any single area. You don’t need to master every tool or understand every algorithm. Focus on being resourceful and creative with what’s available, and maintain enough depth in your core expertise to make good judgment calls.
Collaborate whenever possible, but don’t be afraid to work independently when you need to move fast. Generalists thrive when they can bounce ideas off specialists, but they also shine when they can prototype solutions without waiting for committee approval.
Where all of this madness is headed
I’m pretty sure that “AI generalist” is going to become as common a job title as “software engineer” within the next few years. Startups are already posting positions for “AI Engineers” and “AI Product Associates” where they explicitly want breadth over depth, plus the ability to work autonomously across traditional departmental boundaries.
This doesn’t mean that specialists are gonna go the way of the dinosaur. Someone still needs to build the state-of-the-art models and push the boundaries of what is possible, and the toilet doesn’t unclog itself, amma right? But most companies can’t afford to hire nothing but PhDs, and they don’t need cutting-edge research for every problem. They need practical people who can keep the AI systems running, understand business context, and build solutions that create measurable value.
The magic starts when human experience meets AI capability. The tools can generate cost analyses, suggest architectures, and create prototypes, but you still need business understanding to know which solutions are worth implementing. You need organizational awareness to navigate change management. You need communication skills to help others understand and adopt new approaches.
And under these circumstances, with your level of commitment, and with the right AI tools in your rucksack, you have become a lethal weapon.
Well, at least lethal to the competition that is. . .
If you’re feeling overwhelmed by all this, welcome to the club. I felt the same way when I realized my carefully cultivated expertise in one particular domain might not be enough anymore. But then I reframed the whole situation, and I became an AI generalist who adds value across multiple dimensions.
So here’s my challenge for you. . .
When the next AI tool launches tomorrow (and there will definitely be one – that is if the bubble doesn’t’ pop), think about how it could reshape your entire approach to work. Could you use it for research? Process analysis? Prototyping? Communication? The people who choose comprehensive experimentation over narrow specialization are going to be the ones defining what work looks like for the rest of this decade.
And I think that sounds like way more fun than staying in your lane.
Signing off with writing cramps,
Sorry Marc Drees – my short copy skills are non-existent
Marco
I build AI by day and warn about it by night. I call it job security. Big Tech keeps inflating its promises, and I just bring the pins.
Think a friend would enjoy this too? Share the newsletter and let them join the conversation. LinkedIn, Google and the AI engines appreciates your likes by making my articles available to more readers.
To keep you doomscrolling 👇
- The chicken that refused to die | LinkedIn
- Vibe Coding is gonna spawn the most braindead software generation ever | LinkedIn
- The funniest comments ever left in source code | LinkedIn
- Today’s tech circus is a 1920s Wall Street reboot | LinkedIn
- The Sloppiverse is here, and what are the consequences for writing and speaking? | LinkedIn
- Why we’re all writing like shit on purpose now | LinkedIn
- Creating a hero’s journey for your AI-strategy | LinkedIn
- How corporations fell in love with AI, woke up with ashes, and called It ‘innovation’ | LinkedIn
- The corporate body snatchers | LinkedIn
- Screw your AI witch hunt, you illiterate gits | LinkedIn
- AI perverts make millions undressing your daughters for fun and profit | LinkedIn
- The state of tech-jobs 2025 | LinkedIn
- Meet the sloppers who can’t function without asking AI everything | LinkedIn
Leave a Reply