3 underrated, yet powerful AI skills
Summary
As AI becomes a significant part of our work lives, which competencies differentiate the average business user from another? I argue there are three powerful but game-changing skills that you haven’t thought of in this light.
Clear writing
Thoughtful delegation
Deep, immersive work
Much of the conversation about AI tends to focus on its technicality — the capabilities of the latest foundational models, how many GPUs trained it, the number of tokens it can process, the context window, model fine-tuning, etc. Of course, each technological advancement represents a significant leap in the outcomes AI can achieve. Still, when the rubber hits the road, many AI outcomes depend on a human in the loop — a human who isn’t a data scientist, a developer, or even a technologist. In other words, a typical business user.
Indeed, agents and AGI might complete many tasks autonomously in the future. Still, until humans are around the world of work, most of us will have to work with AI applications. Over the last couple of years of using AI in business settings, I’ve realised that three underrated but powerful skills set apart the people who are successful with AI from those who can’t yet leverage these new tools effectively.
First, we must write clearly
Writing is the number one distributed work superpower. If you can write clearly, you can work asynchronously, and that gives you time away from unnecessary meetings, so you get more out of your 40-hour work week. As simple as that.
It turns out that writing is also an essential AI skill. How do you interact with ChatGPT, for example? Through writing? What about Midjourney, Firefly, Perplexity or Claude - we’re still writing! In fact, the better you write your instructions for any AI tool, the more likely you are to get your desired results in a single shot. The poorer your writing, the more you’ll encounter frustrating back-and-forths with the current crop of AI tools.
My colleague David John sent a message this morning that echoes my sentiments. It’s so eloquent that I must share his thoughts with you.
The biggest win so far has been using GenAI to edit and tighten up my writing. I’ve always cared about getting my wording just right, but that usually meant spending way too long on it. Now, with the right prompts, I’m getting that polished end result in just a few quick iterations.
What’s been even more useful is how much it’s helping me translate tech into human. Being new to the industry, I’ve had to swim through a sea of jargon, product names and words that mean completely different things here than in the outside world. GenAI has made it so much easier to cut through all that and not only understand what’s going on but have it stick.
But the most unexpected win? I actually think it’s making me better at thinking🤪... Writing things out, tweaking them and looking at different angles has helped me be clearer, challenge my own assumptions and spot gaps I would’ve missed. It’s a bit like debating with yourself, just without the weird looks from colleagues.
As someone rightly said, “Good writing is clear thinking made visible.” Each of us has different ways to prompt AI, some more effective than others. When getting to terms with Midjourney and Firefly, I learned to use technical language to describe the artwork, such as neo-pop, double exposure, f-stop for photorealistic outputs, or the style of a famous director, artist, or studio, to mimic their work. When using chatbots, I often use prompts that include context about who the AI is and who I am, a clear challenge and a description of how I’d like my outputs. The diagram below illustrates this approach.
My typical prompt structure, for chatbots
The key is to find our own practical and precise ways to communicate our needs to AI that we can rinse and repeat. Clear writing, therefore, is a foundational skill to facilitate such communication.
Second, practise thoughtful delegation
At Thoughtworks, we use Glean as our work AI platform. While it’s the industry-leading enterprise search platform, one of its most incredible features is its prompt library. The idea is simple. If you prompt the AI to complete a task for you and believe that prompt is reusable, you can save it into the library so your colleagues can benefit from it and simplify or speed up their work.
Last month, Glean did one better. They introduced advanced prompts that allow you to chain multiple tasks together to achieve an overarching goal with AI. Here are some activities you can instruct Glean to complete within these advanced prompts.
Read content from specific files
Search for colleagues
Search your company’s knowledge base
Reason or analyse on your behalf
Search the web
Search for experts
Generate a response
Glean’s prompt engine allows you to combine many simple tasks to produce rich outputs
As fascinating as this capability is, the power of Glean’s advanced prompting lies in its ability to describe work as a series of micro-tasks, where one or many outputs lead to a complete artefact. For example, consider how a salesperson analyses requests for proposals (RFPs) from prospective clients.
They extract several bits of information from the RFP document.
Next, they look for the client’s history with the company on their CRM platform.
The company’s knowledge base gives them helpful information to craft their proposal.
Finally, the salesperson determines how they can create a winning response to the RFP.
For AI to effectively perform this analysis job, a salesperson must be able to decompose the work into its constituent tasks. This act of decomposition is similar to delegating tasks to a human being.
My friend (and colleague) Nag describes Glean as “a smart intern with endless patience and boundless energy.” I agree with that description. To stretch that metaphor further, I’d say that if the current state of AI describes such “interns,” then people who can delegate effectively will get more out of their interns than people who don’t take the time to decompose and explain their work.
And finally, make time for deep work
Many of us dream that AI at work will eliminate the drudgery and toil and leave us with time for joyful and intellectually stimulating activities, like in the Venn diagram below.
Reduce the toilsome work, free up time for valuable work
But generative AI tools like ChatGPT and Copilot aren’t productivity fairies — they’re cognitive mirrors. A recent Microsoft study reveals a paradox: the smarter our AI becomes, the more vulnerable our critical thinking muscles grow to atrophy. When workers lean too heavily on these tools, they trade the friction of deep problem-solving for the illusion of efficiency. The research shows we’re outsourcing not just tasks, but thinking itself. The algorithms do the heavy lifting while our brains hover in cognitive autopilot.
Here’s the twist, though: the study also illuminates a path forward. Workers who paired AI with intentional, deep engagement — those who treated outputs as rough drafts rather than final answers — maintained sharper judgment. They became editors rather than passive consumers, applying their focus to verify facts, refine ideas, and steward quality. Think of a chef using a food processor to chop onions faster but still tasting the soup!
There’s something to say about wielding AI as a sparring partner, not a crutch. As a bright intern, but not a full-blown employee. Working this way means reinvesting the time we earn from delegating toilsome work, into deep, immersive work blocks. Timeblock yourself to build on what AI throws at you. Critique, improve and enhance AI outputs using your style and creativity. Use AI to get a leg up but not to dull your intuition and creativity. Remember, if everyone can produce with AI, then the people who retain differentiation are the ones who can layer human creativity on top of generated outputs.
So, as we integrate more AI into our modern workflows, there is a case to refocus on some fundamental competencies.
The point is to look beyond the chatbot format's back-and-forth interactions and think of ourselves as designers of our new work experience. Writing allows you to describe how AI will assist you. Delegation helps you decompose large pieces of toilsome work without the pain of messy chats. And practicing deep work, strengthens your creative differentiation.
Those three underrated skills are also game-changing AI skills. You may have more to add to the list. Let me know in the comments. I’d love to hear your thoughts.