Written By: Alec Violette
How exactly do you replace an enterprise-level file management system with a team of three developers? Trick question—you don’t. What you can do is create a lean, tailored version that fits your client perfectly. That was our idea when we set out to build a functional, scalable alternative to a well-known full-scale Enterprise Content Management tool (ECM), all while demonstrating our ability to execute a fully custom-coded solution from scratch. And if we could prove that AI can accelerate modern development in the process? Even better.
At srcLogic, most of our work happens in low-code environments, but this project was different. There were no out-of-the-box features or guardrails—just an AWS login and a clear mission. We weren’t aiming to replicate every feature of a large ECM; we were focused on delivering exactly what our client needed: storage, organization, searching, and security. No intricate interface filled with unnecessary bells and whistles—just the basics executed at maximum efficiency.
Of course, knowing what to build is one thing—figuring out how to build it is another. We embraced a fully serverless architecture, relying on cloud storage, event-driven workflows, and a distributed search index. In theory, this meant automatic scaling, minimal maintenance, and no wasted resources. In practice, it meant a steep learning curve and plenty of on-the-fly problem solving.
We didn’t have every detail planned out in advance; this was uncharted territory. None of us had built a system like this before, and while documentation is invaluable, sometimes you need a direct answer without wading through a 20-page spec. That’s where AI became an unexpected ally. Tools like ChatGPT and Copilot weren’t writing our code for us, but they saved us countless hours of frustration—whether it was debugging a perplexing permissions issue, structuring our database calls, or determining the best way to handle event-driven workflows. AI provided the push we needed when roadblocks appeared.
One of the biggest challenges we encountered was search. Our first instinct was to use OpenSearch, as it’s built for this kind of task. However, that decision came with its own hurdle: we had to learn it from scratch. We dove into setting it up, structuring queries, and making it work efficiently in our serverless environment. It was trial by fire, but with enough research—and some AI-powered nudges in the right direction—we got search running smoothly. And in doing so, we essentially built a full-text search engine from the ground up.
And of course, we couldn’t overlook the importance of user experience. What’s a modern file manager without dark mode? In today’s software landscape, dark mode isn’t just a trendy feature—it’s a user-centric necessity that enhances usability.
In conclusion, this project exemplifies our approach at srcLogic: delivering exactly what our clients need, even if it means venturing into new technical territory. We’re continuously learning through trial and error while delivering great value, and we pride ourselves on building solutions that fit our clients perfectly—not just the most elaborate, expensive option on the market.