June Update
Hey all–it’s been a super busy couple of months at Adalo, and we’re absolutely cooking with new developments and system improvements. After significantly increasing the scope of delivering Adalo 3.0 (Adalo 2.0 and 3x the speed) we’re approaching the finish line! For us to declare Adalo 3.0 done (and do justice to the many makers that asked for improved speed), we decided that we had to deliver the following:
- Put all the old database infrastructure in the trash and make the switch to much faster, scalable infrastructure (I called it the Ferrari V12 engine) - done Dec 2024
- Overhaul the 100+ backend systems that were deprecated, overly complex and inefficient - majority completed by Jan 25, but final systems done last week
- Review Adalo’s Builder codebase and rip out any bloated and inefficient code - done by Feb 25
- Optimize caching, especially around images and rendering - done by March 25
- Add database indexing - see the X-Ray update below!
- Upgrade React Native across the board (this is done, waiting on component developers to fix their components and we can roll it out!)
- Refactoring Adalo’s code to stop pulling in lots of unnecessary things that you don’t need, especially on the first screen load. This update is underway, and thanks to AI making a significant impact on development velocity, I’m happy to report that we’re making very fast progress!
We want to deliver Adalo 3.0 before the end of September!
But we’re not only doing speed improvements!
We’re reaching top gear in terms of productivity with our team (which has continued to expand significantly). In fact, our current bottleneck is code reviews–as in we have so many development pushes that it’s hard to review it all fast enough! !
We want to call out a few highlights of what we’ve done, and then provide some commentary on what’s next.
What’s done (in addition to Adalo 3.0 work):
- Adalo X-Ray (beta) - AI that will look through your app and tell you exactly what is affecting your performance! Currently with 10% of makers!
- DreamFactory legacy (non-API) systems integrations (beta) - successful integrations of old and powerful systems into Adalo, meaning your app can ingest anything now!
- The foundational basis for AI prompt engineering, and many other AI-related offerings.
- A score of backend improvements, continuing on delivering improved speed and stability.
What’s next:
- Phased rollouts of the above to 100% of the community. We’re no longer pushing to everyone simultaneously to reduce any risk of performance issues.
- Adalo Function Hub (alphabeta) - a way to no longer need Make or Zapier for many purposes
For free!
- An Experts retest.
- AI prompt engineering, and a series of AI-based components.
- AI Chat - this is already in our staging environments, but we want to make sure its accuracy rate is improved before shipping it.
- A super secret, super cool project!
What’s done (including Adalo 3.0 work):
We’ve found that in the absence of guardrails for how to build apps, Adalo’s builder flexibility has meant some Makers build apps with inefficient flows or data relationships, restricting their app speed.
To address this, we are happy to announce a new performance auditing feature called Adalo X-Ray. X-Ray can perform a real-time analysis of your application, providing feedback regarding:
- The presence of nested lists and their location within specific screens.
- Overall screen complexity.
- The presence of orphaned screens, which unnecessarily increase the overall application size.
- A database table summary, including record counts and relative table sizes.
- Database performance metrics, including among other data tables and index cache hit ratio.
Check out the documentation here!
Applications associated with databases containing tables having more than 10,000 records can be submitted for a special optimization audit. The below screenshot depicts the table and index cache hit ratio data for a mobile app.
For the past several weeks X-Ray has been available in limited beta and is currently being rolled out to a larger audience.
We’ve done a very heavy amount of software upgrades, which rolls in a number of performance and security improvements. This upgraded framework has been tested extensively on our staging servers until our team was unable to break anything!
Currently, we are waiting on our component developers to upgrade their components to ensure compatibility. This process is being overseen by Michael Heck, the (former) founder of No Code Monkey, so we’re appreciating his experience on ensuring the upgrades don’t break anything for any maker!
The upgrades will come with a toggle, so if you wish to run older builds on older React Native versions, you will be able to do that.
Adalo is deployed into a surprisingly large number of corporations across the US, Europe and Japan because of its easy and fast way to spin up web and mobile apps. DreamFactory has enabled many of the older and more awkward systems used by these companies to sync beautifully with Adalo. Examples include Microsoft SQL Server (which comes with no API!), IBM DB2, proprietary ERP systems and AS400-based programs.
This remains in beta (ask if you would like access) but will form the basis of Adalo’s enterprise offering in the months ahead.
When we stepped into Adalo in September 2024, we set out to migrate ~140 services, including the entire structure of databases, to new services. By the close of 2024, we had migrated all but 4 of the services on our list. This last month, we finally completed migrating the last 4 services. It was an extraordinary amount of work considering we manage approximately 3 million PostgreSQL databases. While most Makers won’t see any change, the job of maintaining and running those databases (most of them on forever free plans) is made easier for us, which in turn makes it easier to improve the code base without breaking things. Shoutout to our amazing engineering team!
We have also been improving infrastructure monitoring across the board, and have deployed a new series of Prometheus dashboards to enable this monitoring. We intend to get this monitoring into Makers’ hands, but that will likely be a downstream priority.
Thanks to Maker feedback we were made keenly aware that our rush to release new features and bug fixes was at times counterproductive, and occasionally resulted in unexpected errors. In response, we have implemented a more rigorous and progressive production deployment sequence which is described in this blog post:
What’s Next?
Our team has been working on some major initiatives that will see the light of day in Q3, including:
-
Phased rollouts of the above to 100% of the community. We’re no longer pushing to everyone at risk of slowing down apps.
-
Adalo Functions Hub
- It’s always been a challenge to manipulate data inside an Adalo app without relying on third-party tools such as Zapier. For instance, it is not possible to natively convert a string to lowercase or sort an array of numbers. We are very happy to announce that a new feature called Function Hub will soon offer Makers dozens, and eventually hundreds, of API endpoints for manipulating numbers, strings, dates, times, and more.
- Function Hub is currently in an early stage of development, with the goal of releasing it to a wider audience in Q3. If you would like to be involved in shaping the development of this new feature please email Adalo CTO Jason Gilmore at jason.gilmore@adalo.com.
- (Also, one of our teammates said we should call this ‘Magic Kitchen’. We’d love any comments if you would prefer to call this Function Hub or Magic Kitchen!)
- An Expert retest.
- It has come to our attention that many of the Experts listed on the Adalo marketplace are no longer there, or have been providing unsatisfactory performance.
- We will soon be issuing a retest for Experts to complete, which will require the use of X-Ray to help deliver a fast remake of an app that is specified by our team. The app will be the same for all Experts, and in the event that the Expert is unable to build the app with good performance speed (or they don’t participate in the retest), then they will be delisted from our Marketplace.
- Experts that don’t build an app that is fast will be offered further training from our team. The goal is to ensure Adalo customers continue to get high-quality outcomes from their Experts.
- AI prompt engineering, and a series of AI-based components.
- More to follow soon.
- AI Chat
- This feature has now undergone four rounds of implementation due to both cost and quality considerations and is almost ready for production. We see this as a powerful feature to help Makers quickly move past any sticking points in their build. As soon as we are confident in the accuracy of this feature, it will be available to all users for free.
A super secret, super cool project!