Building my first B2B saas app

A friend of mine owns a concrete company, and one day he showed me how they managed their truck fleet. Almost everything was done manually. Maintenance records were scattered across Excel spreadsheets, with no scheduling or preventive maintenance routine in place. After seeing how much time they spent managing everything by hand, I told him I could build a piece of software to help automate the process.

At first, I thought the solution was simple: build a web application where users could update the status of each truck every day. After spending time with the company, I realized the hardest part wasn’t building the software. It was designing something that fit naturally into the way people already worked.

Discovering an Existing Industry

As I researched the problem, I discovered an entire category of software called CMMS (Computerized Maintenance Management Systems). There were already many mature solutions on the market, which initially made me question whether I was solving a problem that had already been solved.

My best bet was trying to make the software as easy to use as possible, so I thought that an integration with WhatsApp was a good idea, since it's the most used software between mechanics and drivers for their daily activities.

II wanted to transform WhatsApp conversations into structured data. Drivers and mechanics could create service orders, report maintenance issues, or send voice messages directly through WhatsApp. AI, combined with rule-based parsing, would extract the relevant information and automatically update the platform, giving fleet managers a real-time view of vehicle status, maintenance history, and open service orders.

The First Version

Most of the development happened after the first release. Every week, we collected feedback from the people using the platform, discussed their biggest pain points, and shipped improvements based on those conversations.

One suggestion that came directly from drivers and managers was to identify vehicles by their license plate instead of selecting them from a list. A message like “QCZ0I01 has an engine problem” would automatically identify the truck, extract the reported issue, and create a work order for that specific truck.

What I Learned

Software can’t organize a company on its own. Before introducing new tools, there needs to be a clear operational process, consistent record-keeping, and a team committed to following it. Without that foundation, even the best software becomes another system that people ignore.

Building CIME showed me that technology works best when it reinforces good operational practices rather than trying to create them. Once the company had a more organized way of managing maintenance, service orders, and inspections, the platform became a natural extension of those processes instead of another task for employees.

Besides the full-stack and software architecture experience I gained while building CIME, I learned something even more valuable. When I started, I thought I was building fleet management software. Looking back, I was really learning how operational software gets adopted inside traditional businesses.