We started in 2017 with almost nothing. Here is what nine years taught us.

In 2017, I started Elegant IT Limited from Dhaka. We were a small team, honestly very small, and most of our early clients were local businesses who needed a logo, a website, or a basic design. Nothing glamorous. Just work that needed to get done.

Nine years later we have worked with clients in the US, UK, Japan, Hungary, Thailand, and more. We have shipped products for healthcare companies, logistics platforms, fintech startups, and SaaS founders. We have presented at Japan IT Week in Tokyo, the EQUALS Summit in Hungary, and CeBIT ASEAN in Bangkok.

But this post is not about the highlights. This is about what actually made us better — the things we got wrong for years before we figured them out.

The first problem: we were too eager to say yes

Early on, when a client came with a project, we said yes to everything. Any scope. Any timeline. Any budget. We were hungry for work and we did not want to lose a client by asking difficult questions.

This caused us a lot of pain. Projects ran over time. Clients were unhappy because what we delivered was not what they had in their head, and we had never asked the right questions to understand what that was. We were doing the work, but we were not doing the right work.

The fix was simple but took us a long time to accept: spend the first week on discovery. No design, no code, just conversation. What are you building, who is using it, what does success look like in three months? Write it down. Agree on it. Then build.

The most expensive thing you can do in a software project is build the wrong thing very well. We learned this the hard way, more than once.

The second problem: we underestimated our own work

For a long time, we quoted low. We thought being from Bangladesh meant we had to be cheaper than everyone else to win international clients. And honestly, some clients came to us only because of that.

But the clients who paid us less almost always valued the work less too. They pushed more, they respected our process less, and they were harder to work with. The clients who paid a fair price, even when it was higher than they initially expected, were the ones who trusted our judgment, followed our process, and ended up with better products.

We stopped competing on price. We started competing on process, speed, and quality. That changed who we attracted.

What actually works: a fixed process, a fixed timeline

Around 2021 we moved to a model where every project follows the same six-week structure. Discovery in week one. UX flows in week two. Visual design in week three. Engineering in weeks four and five. Launch and handoff in week six.

A lot of clients pushed back at first. Can you really build something real in six weeks? Yes, if the scope is clear. The constraint is the point. It forces everyone — us and the client — to decide what actually matters and what can wait for version two.

Why we moved into AI and what we actually learned from it

We added an AI practice in 2023. Not because it was trendy, but because one of our healthcare clients had a scheduling problem that could not be solved with traditional software. They had six hospitals, hundreds of physicians, and a scheduling team spending half their week manually resolving conflicts.

We built an agent pipeline that automated most of that. It worked. The client went from 14 hours of manual scheduling per week to around two. After that we knew this was something we wanted to build seriously. Not chatbots, not wrappers, but real production AI that changes how a product works.

What is next

We have relaunched with a clearer focus: product design, AI development, full-stack engineering, and animation. The same team, the same process, but better positioned for what founders actually need right now.

If you are building something, an MVP, a platform, a product that needs AI built into it from the start, we would like to talk. Fifteen minutes is enough to tell you honestly whether we are the right fit.