luthiraabeykoon

I came across this project recently and thought it was pretty interesting. Remotion is an open source framework that lets you create videos using React components instead of a traditional video editor.

You define scenes, animations, audio, and transitions in code, and it renders to formats like MP4, WebM, or GIF. Since it’s React-based, it works well for data-driven or templated videos, automated content generation, and anything you’d normally struggle to do in a timeline UI.

Feels especially useful if you’re already comfortable with modern frontend tooling and want reproducible, programmatic video output.

I’ve been working on an open-source project called Julie (desktop AI assistant, local-first, agentic computer use). It started as a personal tool and has grown enough that I’m considering turning it into a real product. I have a ton more features i didnt push yet, including bug fixes and determinisim in the agent behaviour.

I have about $5k I can realistically put toward this and I think it could be sustainable if I can get a small but real group of users.

I’m trying to sanity check a few things before committing:

Is this the kind of thing people actually pay for, or does it stay a hobby/open-source tool?

What would you focus on first to validate demand without burning the budget?

Any advice on early distribution for dev tools like this?

i use qwen3-8B for text and qwen3-vl-4B for vision

Good question. Local LLMs are already “good enough” for most in-context work: short-to-medium writing, refactors, reasoning over what’s on screen, and multi-step agent plans where latency and privacy matter more than raw IQ.

I still fall back to remote models for very long-context tasks, heavier code synthesis, or when you want best possible reasoning over large codebases, the goal is to default local, then escalate only when it actually adds value.

One small note: I’m finally at a place where I’m genuinely happy with where this landed, so I’ll probably pause active development for a bit.

That said, I’m excited to see how people use it, and I’ll still be around to answer questions and fix issues if they come up.

Totally fair concern. I wouldn’t expect anyone to give an assistant full, unsupervised access to their machine.

The model here is supervised control by default. Julie can see context and propose actions, but anything state-changing is gated behind explicit approval. You can also run it in read-only or draft-only modes where it helps you think and write, but never clicks or types on your behalf.

Access is permissioned and scoped, not blanket. Screen, keyboard, mouse, filesystem, network, each is a separate capability with clear boundaries and per-app controls. On top of that, actions go through defined protocols rather than free-form behavior, so you know exactly what kinds of operations it’s allowed to perform and under what conditions.

The goal isn’t full access. It’s controlled access with clear protocols, human-in-the-loop execution, and the ability to shut it off instantly if something feels off.

I thought Cluely was actually a solid idea early on, but over time it felt like it drifted toward pricing, friction, and controversy instead of pure usefulness. That got me wondering if I could build the same core idea, but stripped down, open source, and focused purely on productivity.

OpenAI has a desktop GPT app now, which is great, but it still doesn’t really solve the “don’t break my flow” problem. You end up switching tabs, dragging things around, or copy-pasting context. It’s not bad, but it’s not invisible either.

So I built Julie over a weekend (about 2 days) mostly for fun and to see if I actually could. Julie is a lightweight desktop assistant that lives on top of your workspace, sees what you see, listens when you want it to, and responds without forcing you to context-switch. No cheating angle, no gimmicks, just trying to stay in the same mental lane while you work.

It’s fully open source and costs $0. It uses Groq’s high-throughput inference for text, speech, and vision, and runs as a simple desktop app. It’s not literally invisible or magical, just a practical assistant that reduces task switching instead of adding more UI.

This isn’t meant to replace anything or start drama. I mostly wanted to prove that this kind of assistant can exist without paywalls, subscriptions, or hype. If it’s useful to others, that’s a win.

Repo and installers are up here if anyone wants to try it or poke holes in it.

This is a very minimal version so let me know your thoughts :)

a free leetcode for electrical engineers. real interview problems from nvidia, apple, google, qualcomm, amazon and more. built for anyone who actually wants to get better, fast.

if you have any feeback please feel free to reach out, would love to hear more :)