This year marked the fifth year we’ve been building AppMaster!
We had everything this year: awesome achievements and epic failures. It is crucial to regularly check what we’ve accomplished and see if we’re speeding up or stuck in place.
Overall, 2024 was turbulent for AppMaster: we dove into a bunch of overdue changes and got deeply engrossed in them.
In Q1 we rushed to fix the web app designer that went live at the end of 2023. All new projects started using it, but due to architectural mistakes, the web apps just stopped working as they grew—completely freezing the browser. To top it off, our key developer (responsible for the new version of the web designer) quit a few weeks before the fixed version was set to release. By late March, we got a new web designer that “almost” worked, but only in Chrome and only in very limited scenarios. That’s was an epic failure in terms of bad timing and a way to fail. But our team is strong enough to continue working in many bad situations. So we reconfigured our team and started to work.
In Q2 we finally released the new version of the web app designer, and what a relief—it started to run quickly and with minimal issues. We spent the entire quarter fixing bugs throughout the product, which was a big, somewhat tedious job but necessary to address must-have features and poorly functioning elements.
A huge part of our product is invisible to users—our internal infrastructure and various support services. In Q3 we released an updated version of the entire AppMaster Studio web interface. We moved away from separate web apps stitched together with iframes and adopted a more mature microservice architecture (micro frontends) with Module Federation. This lets different teams develop parts of the project as standalone apps, then piece them together in the user’s browser. End users may not notice, but internally it’s a giant leap forward in simplifying dependencies and boosting development speed.
Also at the start of Q3, we had to abruptly stop using Google Cloud because our Google for Startups credits ended and weren’t renewed. Paying their huge fees (even more expensive than AWS in our case!) made no sense. After a few sleepless nights, we migrated everything to AWS and partly onto our own hardware.
Toward the end of the third quarter, our technical debt finally caught us. The codebase had so many problems that features started breaking in batches whenever we made changes. For a while, the entire platform felt like one giant bug. Worse, tons of nearly finished features (90%+) were stuck in our release pipeline due to cross-dependencies blocking their final release. By late Q3, development almost came to a halt.
Q4 became our turning point. We largely stopped developing new features and focused on killing technical debt, reducing overall product complexity, and simplifying our infrastructure. Fewer components and a more compact setup mean cheaper, faster maintenance and growth. In just a couple of months, we removed several backend microservices, carried out a sweeping optimization of all services and infrastructure, and brought our monthly AWS bill down. Me personally had to dive back into coding again in Q4 to rewrite our real-time collaboration server based on CRDT. It had long been a potential point of failure and caused major issues a few times. In two weekends, I rewrote it from scratch—ditching 99% of the old code and turned it into a super-simple stateless service. Before, the collaboration service needed to “understand” which type of document you were opening (web, mobile app, DB schema, etc.). Now it’s fully abstract with no extra upkeep needed.
Despite all the hurdles, 2024 was huge for us:
- We launched our new web app designer
- Added collaboration to the database designer
- Released two versions of our API Protection with encryption and signatures
- Added i18n support into backend and web applications
- Web applications finally got breakpoints to adopt elements for different screen sizes
- Launched regional pricing to let engineers from low-income regions to us our platform
- We made thousands of small improvements and added tons of functionality
- Dramatically optimized our infrastructure to run even faster and more stable
By the end of 2024, we were just 3–4 weeks away of rolling out the new mobile app designer (v4) and real-time collaboration in the logic editor. Our sales department, established in early Q3, finally started bringing in the first large enterprise deals by year-end—still slow, pricey, and struggling, but we’re moving forward. We also had to replace the entire iOS engineering team in 2024 due to slow dev speed and poor results.
Looking ahead to 2025, our plans are enormous. Thank you all for your support!
Semper ad meliora, numquam cedas!