Not sure what kind of model you're interested in, but you may give Duck.ai[0] a try. It doesn't require any signup and let you use gpt-5.4 nano/mini, Haiku 4.5, Mistral Small 4 and Gemma 4 31B.
Desktop app bundled with an ai model and a local-first agent so nothing leaves your machine.
Last week I interviewed non-technical people about their experience with AI agents. Many couldn't even use them at all. Either they didn't want to share private data with ChatGPT or company policy prohibited it.
For those of you working with sensitive files – contracts, client records, financials, HR docs – I'd love to hear how you handle this today: simone [at] breadboards.io
Not right now but I'm actually exploring this idea space — doing user research —. I would be happy to chat with you – or whoever happens to read this and is into this topic – about that. simone [at] breadboards.io
Hey! I'm Simone, a computer programmer from Italy living in Zürich.
I have 12+ years of experience as a full stack developer building web and mobile applications across Europe. I have an extensive startup and agency background, which taught me how to integrate effectively with diverse teams and adapt to fast-moving environments.
I value software architecture design. I have deep expertise in the JavaScript/TypeScript ecosystems and I often explore other languages and environments out of curiosity or when a project needs it. I'm also actively exploring applied AI, from building with LLM agents to tinkering with local models – particularly the smaller ones –. I like to bring a strong eye for clean, minimalist UI design to the products I work on.
Location: Zürich, Switzerland
Remote: Yes
Willing to relocate: No
Technologies: JavaScript, TypeScript, nginx, macOS, Alpine Linux, Node, Java, Swift, SQL and a quick learner
Résumé/CV: https://simquat.com
Email: mail[at]simquat.com
I don't see low-code and LLMs really competing with each other. It may be true for quick throwaway projects, but they solve different problems and can potentially be used together.
With low-code, users operate at a high abstraction level and get a deterministic output. You're fully in control of what the result will be. With LLMs you operate at an even higher abstraction level – just a prompt in plain English – but the output is non-deterministic.
So if you want fine control, you need to check line by line what it produced. I think it gets interesting if LLMs generate low-code instead of code. Users get the speed advantage of AI generation, but they can still understand and control what the software is doing.
I would recommend Scratch, the blocks prevent syntax errors, plus there's a simple editor for drawing sprites and backgrounds — my niece used to play with it paired with the book "Scratch programming playground" if I recall correct —.
I'm working on a visual app builder, Breadboard[0], we're not targeting education neither games, but here an example of a tic tac toe game made with it.
Looks quite cool and I'd like to give a try. What is the main use case for compiling code to shortcuts? I ask because I'm working on a tool[0] that in a way does the opposite.
What you're doing is visual programming. On its own there isn't anything wrong with it. However, specifically with Shortcuts it's not very pleasant for anything complex.
I had a full garden automation running on shortcuts, but it was extreme hard to maintain and improve due to "editor" being so bare bones.
Thanks for sharing. By the editor being bare bones do you mean some missing feature might change your mind about using it, or do you find the text-based editor much more comfortable?
I was talking specifically about Apple Shortcuts. Once you have anything complex, you're going to need code reuse and its a sad story with shortcuts as provided by Apple.
I'm working on Breadboard[0], a modern HyperCard for web apps.
We recently added an AI integration, starting with a UI agent. We're experimenting with a BYOK approach so anyone can try the assistant in the playground[1] without signing in while keeping it sustainable for us. Currently the AI integration connect to Gemini.
A logic agent is in progress, it's a bit trickier because it needs to work with Breadboard's visual-stacked-instructions language based on Hyperscript.
[0] https://duck.ai/