A friend of ours worked in the finance team of one of Pakistan’s biggest packaging companies. He told us once about a reconciliation process that basically ran over email: a single Excel file passed between five or six people, each one opening it, editing their section, and sending it to the next person, because the data needed that many hands on it before it was done. Listening to him describe it, we kept thinking a shared Google Sheet would have solved most of this in an afternoon. Same file, everyone editing at once, no inbox full of six different versions of the same spreadsheet. The fix wasn’t even hard. Nobody there had just stopped to ask why the process worked this way to begin with.
Once we started looking for that pattern, we found it almost everywhere. Accounting came up a lot, probably because it’s one of the easier things in a business to automate, and yet at most companies we looked at, it was still mostly people doing it by hand. It usually wasn’t about money or bad software. Most of the time, people had quietly merged “AI” and “automation” into one vague idea they didn’t fully trust, something that felt like either a fad or a threat to their job, and rarely something that could make a Tuesday less annoying. So a task that could take twenty minutes stayed a two hour task, week after week.
What got us was who this was happening to. These weren’t companies scraping by on outdated tools because they couldn’t afford better. A lot of them had healthy margins and real budgets, the kind of firms that could have grown those margins further just by fixing the small, boring friction inside their own teams. Nobody had looked closely enough to notice it was even there.
That’s what got us started. We worked with small businesses first, on purpose, well before we felt ready to sit across from a CFO and tell them how to run part of their operation. Working with smaller clients taught us where automation actually helps and where it just creates a new mess nobody wanted. We wanted that judgment before we tried to sell anyone on it.
We’ve come to think that judgment is the actual scarce part of this work. Engineers can build almost anything you ask them to. Fewer people can walk into a business, figure out which process is worth automating and which one should be left alone, and then introduce it in a way that people will actually keep using. People don’t like change, and we’ve stopped treating that as something to argue them out of. It’s just something to plan around. If a system isn’t fast, obvious, and trustworthy from the moment someone opens it, they’ll quietly stop using it, no matter how well it was built. That’s really the whole job: the technology, and the people who have to live with it.
It’s rarely hidden. Most businesses have at least one process nobody has questioned in years, mostly because it’s always been done that way and it technically still works. The first few weeks of any engagement are really just us watching for it, sitting with the people actually doing the work instead of reading a process document someone wrote two years ago.
Not everything manual should be automated, and some things that should be aren’t worth the disruption just yet. This is the part we can’t hand off entirely to an engineer, because it takes someone who understands the business well enough to tell which fix actually pays for itself and which one just adds a different kind of complexity.
A system nobody trusts by the end of the first week tends to get quietly abandoned, no matter how well it was engineered underneath. We spend as much time on how something feels to use as we do on whether it works, because for the person opening it every morning, those are really the same question.
We want to mention something that isn’t really about the business side of Gruntflow. We care about education, and if you work at or run an NGO focused on it, we’ll build your system for free. Reach out through the form below and just mention that you’re an NGO working in education. That’s really all we need to hear.