What’s the most ambitious thing you’re working on this year?
For me, it’s building a startup in the events industry. I’ve always felt the market is underserved — there are so many untapped opportunities in how people connect, organize, and create meaningful experiences.
I’d love to hear what you’re working on. Share it in the comments — who knows, maybe someone resonates with your vision, and you end up talking, collaborating, or pushing each other closer to your goals.
Trending on Indie Hackers
At the age of 50, I fulfilled a dream and created a full-fledged CLI application for Linux beginners. Integrating AI into the command line helps you manage your operating system without endless Googling and stress. I found myself here trying to figure out how to share a useful and free product with as many beginners as possible. For me, this is very ambitious (and challenging).
Oooh sounds interesting but how do you handle sensitive commands? Like deleting core programs in Linux. I havent tinkered enough around to abolish my system but I've done enough to make my Ubuntu stuck in bootloader loading screen. Especially many beginner users as you mentioned as your target users. Aren't they vulnerable to execute bad commands. Once an irrevocable small prompt mistake and they'll erase their existence?
Thank you for your comment. It's nice to meet a colleague — I myself have destroyed dozens of Linux systems, and I managed to do it without any help from neural networks. Besides, the documentation clearly states in large print: BE CAREFUL WHEN EXECUTING UNKNOWN COMMANDS (hahaha).
Seriously though, I think having a good assistant right in the command line will give Linux a chance to live a little longer. My virtual machine is still alive after the assistant appeared, although you would be horrified to see what I did to it.
Perhaps I didn't express myself quite correctly; the neural network does not execute commands on its own, it only suggests actions in response to a request, taking into account the context. But you've actually raised a good point, thank you. Perhaps now I'll configure the LLM context to warn about potentially unsafe commands.
Please don't consider this an advertisement, but perhaps you could take a look at the repository and description and give me some more useful advice. At this stage, any help and criticism would be welcome. https://github.com/Vivatist/ai-ebash
Ah yes youre very welcome. But since you ask for criticism here are my thoughts: can i ask did you do your research before you got into development?
Do you have plan for commercialization or sustainability for your project?.
I did a quick research on a tool something like what youre proposing and what youve built: ive already look on your repo and its clearer to me that you're utilizing llm wrappers. Which i found and searched on github for "cli help with ai" and found similar projects heres the top result.
https://github.com/bjarneo/tAI
If you have plan for commercializing this to a small company it needs a lot of polish to be at least a good consumer product for helping users with cli commands and maybe with context knowledge on their data files as well. You have a long way ahead to become license-able to enterprises and integrate on linux distros if thats how you imagined this at the end. But anyhow,That’s a long road, but exciting too. Good luck on your journey, and I hope this perspective helps.
Yes, I understand what you mean. It's true that I didn't study the market before creating it. The story is quite mundane: my son was struggling with Linux settings at university, so I threw together an AI assistant for him in a couple of days. When I realized how useful the tool was, I decided to refine it and share it with the community. It was only then that I saw that there were other offerings, including the repository you mentioned.
However, my product is fundamentally different from the others in that it does not have strict limits on LLM responses, which restrict functionality to the simple execution of a single suggested command. Therefore, its responses are more suitable for beginners as they contain not only Linux commands, but also comments on them and alternative commands. Compare the output of the command from the tai example given in the repository and my product: https://thumbsnap.com/i/YT3mnVYp.jpg
At the same time, as in competing solutions, if there is a block of code (or several) in the LLM response, it is possible to simply run it without leaving the dialogue with the neural network.
Other features:
You are right that the product still seems raw, I am working on it, but there is still a long way to go. Another problem (and your answer confirms this) is cumbersome, ill-conceived positioning. I have no plans to commercialize the product, but it would be nice if it found its user and my work did not go to waste. Weak positioning undoubtedly hinders this.
Thank you for your response. I welcome any additional ideas, comments, and questions.
My most ambitious activity this year is my first startup. I created a service that allows you to use your own PC as a server for running web applications. I developed the MVP, and now I'm looking for a co-founder to help with product vision and marketing.
I am building a members only exclusive party app.
AI Video creator