The Context Trap: Why Software is Never a Solution
How Human Subject Matter Expertise Governs Business Results
Nicolaes Maes. The Idle Servant. 1655. Oil on canvas. The National Gallery, London. Accession no. NG1221.
Recently during an AI discussion on Discord, I realised that many people focus more on what tool they use, instead of on the problem itself.
Do you really think ChatGPT or Claude will solve your workflow issue?
Software is not the answer here, but rather how you perceive the problem you have to solve.
In a world where prompt engineering has become a basic skill, it’s no longer a question of how, but rather why.
Defining a clear problem is the first step to getting a result that is better than mediocre. Now, being “mediocre” or “good enough” is simply not enough if there is no actual problem analysis.
AI tools become useless without full business context. Yet, you would run out of time trying to share all the nuances and political context—the stakeholder dynamics and sponsor expectations—that won't be free of your own biases anyway.
There is always a part of the work you have to do yourself:
Synthesize and interpret the information.
Both input and output should be human-curated. AI doesn’t know what “good” looks like, especially in heavily regulated industries. Human logic and input are essential if you are looking for results above the average. Otherwise, not only do we start to sound like AI (fluent, polished structure), but entire organizations lose their own voice and culture by abusing AI and blindly copy-pasting whatever the software suggests.
If you have domain expertise, an ethical approach and understand your "why," you can turn raw output into a business result.
How to ensure you provide the right context in 3 steps:
Write down your business goal in one sentence (by yourself).
Identify the constraints (the do’s and don’ts).
Start designing on these strong foundations—using AI as a steer, but you are the captain.
The most irreplaceable skill in 2026 isn't knowing how to just use AI models to save time. This is not enough. It’s knowing what to ask for based on deep business understanding. You have built that through years of expertise. Build on top of it.
The goal isn't to work faster. It’s to work better. And that is a choice only the human can make.