Why Are There No Good Backend Platforms?

This is the tale of our on-premise and microservices AI development platform 1Backend.

Wonder

Shortly after the start of the AI craze, I did what any sensible person would do: together with my brother, who is also an engineer, we dropped about 5 (10?) thousand dollars to set up our own little on-premise AI installation.

It was a wonderfully exciting few months—a kind of joy I hadn't felt in a while when it comes to engineering. You see, although you might not guess it from my age (35), I've been doing engineering for more than 20 years. After a while, you realize that this industry runs in circles. Old becomes new, and new becomes old.

But this AI thing... overhyped? Sure. Often underwhelming? Sure. Still, there’s a spark of magic, a spark of intellect that captivates the whole world. For a moment, we all became children again. Sometimes I feel sad for my son, who missed the magical era of the internet becoming widespread, AI becoming a common talking point. For him, it will be natural and banal, like tap water. But perhaps people who witnessed the adoption of tap water said the same about me, and my son will have his own wonders to behold.

2023: The Beginnings

After setting up our own metal to run AI, we quickly realized that there weren’t many flawlessly working options to run AI models locally. At least, not the way we wanted.
"Seems like we're still early in the AI age!"—I thought to myself.

Realizing this, I did what any sensible person would do: embarked on the journey of writing a few hundred thousand lines of code to create the AI environment that fit my taste—significantly reducing my son's inheritance in the process. Every hour spent hacking on this is an hour not billed. Sorry, son! Daddy is one of the worker bees helping make a post-scarcity society a reality. Maybe the future will have flying cars after all.

After a great deal of hacking, we managed to put the spark into the expensive metal cages. AI was buzzing away happily, answering all our questions tirelessly and privately. Without snitching on us to Big Tech. The genie was in the bottle. Some overheating issues aside, things were good.

There was just one problem.

2024: AI vs. Microservices

Talking to an AI is exciting, but it doesn’t pay the bills. It was time to do what I’ve been doing since the age of 14: start shipping MVPs on top of these AIs. On the frontend, I use Angular—a framework that, despite its flaws, I’ve loved since 2013. But the backend?

At this point, I had to realize: there is no backend platform that makes me happy.

I’ve been building backend platforms all my life. At Hailo (~2013), we were among the first in Europe to jump on the microservices craze after Netflix and Amazon. In 2020, I joined my good friend Asim Aslam in his quest to build the microservices framework/platform. Between those experiences and after, I’ve seen all kinds of backend setups—whether as an employee, contractor, or simply through knowing people who worked on them.

Although Micro contained a lot of novel ideas, it was still Asim's vision. Not that we disagreed—it just wasn’t my painting. And by the time the company received its seed round, the direction was already set. A seemingly small decision that, I think, answers the title question of this post:

The reason there are no good backend platforms yet is that although Micro could have been one, it was too Go-focused and library-heavy.
In other words, it wasn’t language-agnostic enough.

An Incredible Number of New Products

If I can’t find a proper backend platform to build on, chances are a lot of other people can’t either. And with the AI age upon us, the next decade will see an incredible number of new products being developed.

So, taking ideas all the way back from 2013, I set out to extend 1Backend—to add microservice/backend platform development capabilities. What started as a relatively simple on-premise AI tool evolved into a complete, language-agnostic backend development platform.

People in the Java world might call it an "application server." Others might refer to it simply as a backend platform.
I call it a microservices platform—not for buzzword points, but because I think there are so few great microservices platforms that it makes sense to highlight it.

Web frameworks are well-known to everyone—Rails, Laravel, Yii, etc. But microservices? That’s different. The whole point of microservices is that subsystems are separated at the network level. And when you introduce network calls, an entire host of issues arises—from service discovery to container management to an almost infinite number of other challenges.

That’s why 1Backend might seem like a Frankencreature at first sight. The sausage is being made, and you get to watch the process.

It ain’t always pretty!

A Bright Future

While the zeitgeist suggests that anyone can bootstrap a SaaS company in an afternoon without technical chops and apparently hit $300K ARR within a week, my claim—that I built a multi-tenant backend Facebook Pixel product on top of 1Backend in a few hours with about 200 lines of code, now used by a few Facebook pages and e-commerce sites—might seem underwhelming.

Still, I'm rather proud of it, as 1Backend has achieved what I consider the holy grail of microservices: an extendable system where different products built in different languages and stacks can coexist without becoming a tangled mess. A hotbed for prototyping, where new products and features can be launched without breaking existing ones.

This is exactly what our AI age needs, and 1Backend has your back—if you invest a bit of time in studying the documentation.

Or you could just ask an AI to study the documentation for you and implement whatever you want, right?