Putting off AI has a cost
Published · Digidaya
Whenever we talk to companies about AI, the answer tends to be the same: “Interesting, but not yet. We are not ready.” Fair enough. Most AI news sounds like something for big tech firms, not for a mining contractor or a factory in Cikarang. The trouble is that “not yet” is also a decision, and it has a cost. That cost never shows up as a line in the financial statements, so it rarely gets counted.
What AI actually does is simple
Forget robots and jargon. Inside a company, AI pays off fastest in the dull work: reading documents, copying figures, matching records, summarising.
Take recruitment. One vacancy for an operator can bring in hundreds of CVs through email, WhatsApp and job portals. A recruiter goes through them one by one. Someone who applied on day one gets a call the following week, and by then has often taken another offer. AI can read all of those CVs in minutes, rank the closest matches and summarise the interview notes. Who moves on is still the recruiter’s call.
Or take the budget. A site manager wants to know why diesel costs jumped this month. Normally they ask someone to put together a recap and wait a day or two. If the data can be asked directly, in everyday language, the answer comes back on the spot, with a chart per unit.
Engineering is much the same. Reading electrical drawings to build a bill of materials and a quotation can take several days. AI prepares the first draft, and the engineer checks and corrects it. The quote goes out sooner, with fewer missed items.
The pattern is always the same. AI does the tiring part, and people keep the decision. Those three examples are what we built SALVAN, SEKALA and SCHEMAX for.
How to measure it
AI is easiest to justify when you measure it per process. How many days from opening a vacancy to hiring someone. How long from a customer request to a quote sent. How many staff hours go into month-end recaps. How often a wrong entry is only caught later.
Write those numbers down before you start. Without a baseline, it is hard to prove anything to the board.
Work out the cost of waiting yourself
Use your own office’s numbers. Say five administrative staff each spend two hours a day copying data from one system into another. That is ten hours a day, roughly 220 hours a month. Multiply by their hourly cost. That amount is lost every month, and it keeps being lost for as long as the process stays the same.
And that is only the part you can count. The more expensive part is usually invisible: the good candidate who went to another company, the tender lost because the quote was a day late, the cost overrun nobody saw until the books were closed.
One more thing tends to be forgotten. Companies that start early are not just cheaper to run today. Their data gets cleaner, their people get used to it, and the second and third AI projects become much easier. Those who start later have to catch up on the technology and the habits at the same time.
Start small
You do not need a year-long transformation programme. Pick the one process your team complains about most, measure where it stands now, and try AI on it for a few weeks. If the result is clear, move on to the next process. If it is not, you have lost a few weeks, not a year’s budget.
For sensitive data such as employee records, budget figures or project drawings, the AI can run on your company’s own servers. The data does not have to go anywhere. How to choose between on-premise and SaaS.
If you want to know which process in your company would pay off fastest, talk to the Digidaya team. One conversation is usually enough to find the first candidate.