magic VS sakila

I created Hyperlambda (https://hyperlambda.dev), so I spend a lot of time thinking about accidental complexity and implementation surface area. One thing leaks like this keep reminding me is that a surprising amount of software risk comes from packaging and delivery details rather than the main logic people spend all their time reviewing.

I know there's a lot of you guys here cloning magic, and I want to apologise since previously it wasn't possible to run locally due to an initialisation bug of the database.

I've fixed this now, and you can download magic cloud below.

https://github.com/polterguy/magic

What is Magic Cloud? AI-based no-code and low-code code generator and AI agent toolkit :)

Hi guys, gals, and kittens. I've created an Open Source vibe coding platform, and I'd love feedback and feature suggestions for it :)

Repo ==> https://github.com/polterguy/magic

Video demonstrating its capabilities ==> https://www.youtube.com/watch?v=IvoPf6-N3MA

Psst, you can only run it on OSx (M2/M3), in addition to Linux, because I haven't bothered to create binaries for the VSS plugin I'm using in SQLite to deal with embeddings ...

And while we're at it, if you count commits as cognitive work, Magic had 8,007 at the time I closed its source code. Later it's been accumulating some roughly 570 additional commits. 99% of these commits are done by yours truly.

Magic Cloud is the name of our platform. Magic is what allows us to deliver our AI solutions - So writing about our technology's security basically implies writing about Magic Cloud's security.

In the island of Cyprus, I'm the by far most productive GitHub developer with some roughly 12,000 commits. If you check out my username which is "polterguy", you will only see me on 4th place or something - However, this is because GitHub resets commits back to zero if you change your email address, and I really don't give a sjit, so I've not bothered to fix it. If you want to know "my credentials", then look at my work.

Hence, by creating small reusable AI functions, such as we've got the ability to do at AINIRO, any human being can prompt engineer a rich and complex AI app together, allowing his or her users to use natural language to interact with the software.

We already have several clients that are using our AI chatbot for these purposes - However, probably the best use case is our own software system, which is highly complex, and therefor difficult to initially understand. We get a lot of questions about Magic Cloud, and even though I love answering these questions, a lot of questions can be automated by the AI.

I've got an AI chatbot company, and because of that, I have to somehow relate to Google code. Google code is everywhere, and creating a website without using Google code is almost impossible for these reasons. Google Analytics being one reason, and Google reCAPTCHA being another reason. If you've got a website, there's a 99% probability that your site is running some Google code.

Can somebody that's like super smart on React tell me what would be needed to implement this thing in React? The current stuff is Angular, and its code can be found here. What I would need is as follows.

Sorry guy and gals, we just published a new feature in our CRUD generator, and it's basically an AI system that 100% perfectly replaces the human software developer with machine learning algorithms, so you won't be needed anymore at work. Sorry about that, but maybe you could learn how to play the violin ...?

My business is CRUD, literally. In fact, I've got a company around the axiom of CRUD. We're a handful of employees, we've got some VC funding, and we literally "do CRUD" 100% of our time one way or another. Of course, this implies having to think of a lot of other things, such as authentication, authorisation, security, etc - But the basic axiom around which we evolve is that life is too short to having to manually create CRUD apps.

For the record, I would not recommend anyone to stuff 2 billion records into a MySQL database, but it is possible, and Magic handles it just fine. Of course, simply counting 2 billion records takes minutes, and in order to be able to even use it in a Hyperlambda generated CRUD backend you'd have to apply some super aggressive server-side caching constructs, but it is possible as you can clearly see from the following screenshot.

And of course, the code generated by our system is easily several orders of magnitude higher quality, according to every single neutral metric we use to measure quality. In fact, run this code through one of your frontend/Angular developers if you want to assess it from a quality perspective. I guarantee you that he won’t find a single bug in it! Then realize that this code is running here. Try it out for yourself by logging in with admin/admin.