Every manager knows this. You raise a seemingly trivial request with your software supplier: add one field to a form, change how a discount is calculated, build a new report. In your head it's "an afternoon's work". And in reply you get an estimate of three months and tens of thousands of zlotys. Where does this gulf between intuition and reality come from?
Why "small" doesn't mean "fast"
From a business perspective the change is small. From the system's perspective — not necessarily. The classic software development process has its fixed costs, which don't disappear just because the fix is minor.
- Understanding the system — the programmer must first reconstruct how code written years ago by someone else works, often with no documentation.
- The domino effect — one change can drag ten other places with it that need checking so nothing breaks.
- Testing and verification — a solid supplier won't deploy a change without confirming the rest still works, and that takes time.
- The queue — your fix waits until an expensive specialist frees up, and they already have a full calendar.
In other words: you're not paying for the change itself, but for the whole ritual around it. And that's why a "small change" so often costs as much as a big one.
Where the time really goes
If you broke a typical minor-fix project into its parts, the work on the change itself is often a fraction of the whole. The rest is analysing the existing system, manual testing, fixing what the tests caught, and waiting in the queue. It's precisely these stages — tedious, repetitive and dependent on one specific person's availability — that stretch an "afternoon" into a quarter. And it's precisely these stages that lend themselves best to automation.
The ESKOM.AI approach: automating what is time-consuming
At ESKOM.AI we combine engineering experience with automated software development supported by dozens of AI agents. These most labour-intensive, repetitive stages — analysing the code, preparing the change and, above all, the full suite of tests — are largely taken over by automation, which works faster than a human and without fatigue.
- System analysis — what takes a programmer days, AI agents map in hours, delivering a clear picture of the dependencies.
- Automated tests — unit, integration, performance, security and regression tests run on every change, so you know at once whether something broke.
- A shorter queue — since one change requires fewer hours of an expensive specialist, more requests fit into the same time.
A human still guards quality, makes the decisions and is accountable for the result — but no longer loses weeks on things a machine does faster and more accurately.
The concrete: from three months to a few days
The effect is measurable. A fix that was classically estimated at a quarter reaches users in many cases within days. The cost drops to a fraction of the classic rate, because you pay for a real result, not for hours spent on tedious analysis and manual testing. Equally important — the change is safer, because each time it passes through the full set of tests that a quick "manual" fix usually skips. Shorter, cheaper and more reliable all at once.
Check it on your own example
Do you have a list of "small changes" that have waited in the queue for months because they always come out too expensive or too slow? That's the perfect starting point. ESKOM.AI — a company operating since 2025 — will help estimate how much of that backlog can be delivered faster and cheaper than you expect. Write to us and let's arrange a short, free consultation. We'll show you concrete timings and costs before you make any decision.