brainless

I prefer using LLM. But many people will ask what is an LLM and then I use AI and they get it. Unfortunate.

At the same time, LLMs are not a bullshit generator. They do not know the meaning of what they generate but the output is important to us. It is like saying a cooker knows the egg is being boiled. I care about the egg, cooker can do its job without knowing what an egg is. Still very valuable.

Totally agree with the platform approach. More models should be available to be run own own hardware. At least 3rd party cloud provider hardware. But Chinese models have dominated this now.

ChatGPT may not last long unless they figure out something, given the "code red" situation is already in their company.

Very happy to see this since I am building in this domain. We need external and internal context though. I am aiming for always available context for current and related projects, reference projects, documentation, library usage, commands available (npm, python,...), tasks, past prompts, etc. all in one product. My product, nocodo (1), is built by coding agents, Claude Code (Sonnet only) and opencode (Grok Code Fast 1 and GLM 4.6).

I just made a video (2) on how I prompt with Claude Code, ask for research from related projects, build context with multiple documents, then converge into a task document, shared that with another coding agent, opencode (with Grok or GLM) and then review with Claude Code.

nocodo is itself a challenge for me: I do not write or review code line by line. I spend most of the time in this higher level context gathering, planning etc. All these techniques will be integrated and available inside nocodo. I do not use MCPs, and nocodo does not have MCPs.

I do not think plugging into existing coding agents work, not how I am building. I think building full-stack is the way, from prompt to deployed software. Consumers will step away from anything other than planning. The coding agent will be more a planning tool. Everything else will slowly vanish.

Cheers to more folks building here!

1. https://github.com/brainless/nocodo 2. https://youtu.be/Hw4IIAvRTlY

My experience has been the opposite. I came from Python and Typescript and the initial amount of reading and fighting with the compiler was very frustrating but I understood one thing that sets Rust apart - I write code with almost the same level of bugs as a seasoned Rust developer. That is a win for years to come as team grows and software gets old. I will bet on it again and again.

Now I mostly generate code with coding agents and almost everything I create is Rust based - web backend to desktop app. Let the LLM/agent fight with the compiler.

May I add Gliner to this? The original Python version and the Rust version. Fantastic (non LLM) models for entity extraction. There are many others.

I really think using small models for a lot of smell tasks is the best way forward but it's not easy to orchestrate.

  Location: Kolkata, India
  Remote: Yes and willing to travel frequently
  Willing to relocate: No
  Technologies: Agentic development, Go, Rust, TypeScript, etc. Strongly typed stack only. I do not write code by hand anymore, and I have my own agent. You can hire me to build for you or you can build on top and I consult for you.
  Résumé/CV: https://www.linkedin.com/in/brainess, https://brainless.in, https://github.com/brainless
  Email: My name, without spaces or any other characters, at Google's consumer email domain

I am Sumit and I am building https://github.com/brainless/nocodo. I have built this only with coding agents. If you are building rapidly with LLMs, I am very interested to work with you. Use LLMs in every stage of your workflow. Strongly typed languages only. I have been an engineer for 16 years, founder many times and have deep knowledge and experience in early stage product building.

Think of it as Devin AI but self-hosted, headless (use my UI or your), customizable, multi-model, multi-OS (build/debug your apps on Win, Mac, Linux). Enforce policies with hooks. Commercial license at $90K/year with support.

Thanks for sharing.

I did not know of this and I am looking for simple ways to isolate processes for multiple reasons. I am building a coding agent, https://github.com/brainless/nocodo, that runs (headless) on a Linux instance. Generated code is immediately available for demo.

I am new to isolation and not looking for a container based approach. Isolation from a security standpoint but I do not know enough. This approach looks like a great start for me.

This was recently on HN and I think it adds so much value to GPUI: https://github.com/longbridge/gpui-component/

In my coding agent, nocodo (1), I am thinking about using copy on write filesystems for cheaper multi-agent operations. But to be honest git worktree may be good enough for most use cases. nocodo checks existing worktree in the local repo and I will add creation and merge support too.

1. https://github.com/brainless/nocodo

This is interesting to read and very important to me since I am building a coding agent with team collaboration in mind. I used to use Zed daily till the point that I moved away from writing code directly and instead generate all my projects only from prompts.

I think collaboration for people who eventually use the software will be more critical in the era of agentic coding. Project Management will change. We are not waiting for 2 weeks to build prototypes, it gets done in a hour. What does that mean for end users - do they prompt their changes and get access to new software? Who would double-check, would AI reviews be good enough, would AI agents collaborate along with humans in the loop?

There are so many questions not answered. If anyone is keen on having these talks, I would happy to share what I think. Here is what I am building: https://github.com/brainless/nocodo

I want to see a future where end users can prompt their needs, have collaborators in the company to help clear things up and in an hour the feature/issue is tackled and deployed.

That is a valid question.

But that would apply to any app that deals with files like this one does.

This one is open source and we can run some code analysis on it, compile locally, etc. I am not well versed in security checks but I guess you get the idea.

The CLAUDE.md file is right there, so they are probably using agentic coding.

But why does it matter? Does the app not work? I don't have a Mac, can't check.

I am building a coding agent for small businesses. The agent runs on Linux box on own cloud. Desktop and mobile apps to chat with AI models and generate software as needed.

SSH based access with HTTP port forward. Team collaboration, multiple models, git based workflow, test deployment automation, etc.

Very early stage but it now work on its own source code (Bash tool is missing): https://github.com/brainless/nocodo

This is a good point. Should we ask why so many people are still going to ChatGPT? Do the existing systems get so many users interacting with them?

"We are not ourselves when we are fallen down" - hits hard. I really hope this is a calling for folks who will care.

You are right and it gives us an chance to do something about it. We always had data about people who are struggling but now we see how many are trying to reach out for advice or help.

This is not suprising at all. Having gone through therapy a few years back, I would have had a chat with LLMs if I was in a poor mental health situation. There is no other system that is available at scale, 24x7 on my phone.

A chat like this is not a solution though, it is an indicator that our societies have issues is large parts of our population that we are unable to deal with. We are not helping enough people. Topics like mental health are still difficult to discuss in many places. Getting help is much harder.

I do not know what OpenAI and other companies will do about it and I do not expect them to jump in to solve such a complex social issue. But perhaps this inspires other founders who may want to build a company to tackle this at scale. Focusing on help, not profits. This is not easy, but some folks will take such challenges. I choose to believe that.

This looks very good from the screenshots. I will try this as quickly as I can. I have been building with egui and have tried Iced, Slint and Makepad.

I built (agentic coded) a stocks viewer app for Indian stock market data: https://github.com/brainless/Indistocks. It was a fantastic experience as to how easily I could build a GUI app.

My main product also uses egui: https://github.com/brainless/nocodo. It used to have a web app frontend and I moved to desktop app after the experiment with Indistocks. The experience has been really good, also coded with agents.

Desktop apps are fun and even on my somewhat old and slow laptop (i5 8th gen, 16 GB RAM, 2GB nvidia dedicated graphics), they are so much faster than web apps on Chrome (on Linux). I want desktop apps to make a big comeback, we could use so many old devices.

It is not easy to support multiple providers at each layer of our tech stack but if we do not then a few players become easier to pick and then they monopolize.

I am trying to stay vendor neutral with my own coding agent (1). To approach this, I created a desktop app that connects to coding agent running on either my own infra or yours (local or your cloud server). Desktop app and coding agent are separate binaries.

If you host on your own infra then you can bring your own AI provider too. Similarly, I want to give choice for git host. Right now I am targetting GitHub but I want to add Gitlab soon after MVP. All this has made the path to my MVP longer but I see a clear long-term aim for myself - we should have choices.

1. https://github.com/brainless/nocodo

Thank you. Let's see if I can get nocodo to be a power tool for everyone who wants a coding agent.

I vibe coded a desktop app to view Indian stocks: https://github.com/brainless/Indistocks

This is built with Rust, egui and SQLite3. The app has a downloader for NSE India reports. These are the daily end of day stock prices. Out of the box the app is really fast, which is expected but still surprises me. I am going to work on improving the stocks chart. I also want to add an AI assisted stocks analyst. Since all the stocks data is on the SQLite3 DB, I should be able to express my stocks screening ideas as plain text and let an LLM generate the SQL and show me in my data grid.

It was really interesting to generate it within 3 days. I had just a few places where I had to copy from app (std) log and paste into my prompt. Most of the time just describing the features was enough. Rust compiler did most of the heavy lifting. I have used a mix of Claude Code and OpenCode (with either GLM 4.5 or Grok Code Fast 1).

I have been generating full-stack web apps. I built and launched https://github.com/brainless/letsorder (https://letsorder.app/). Building full-stack web apps is basically building 2 apps (at a minimum) so desktop apps are way better it seems.

In the long-term, I plan to build and help others generated apps. I am building a vibe coding platform (https://github.com/brainless/nocodo). I have a couple early stage founders I consult for who take my guidance to generate their products (web and mobile apps + backend).

AI coding for entire business teams, no tech knowledge needed.

https://github.com/brainless/nocodo

Self-hosted, multiple models, bring your own keys and subscriptions, unlimited projects, tasks, web based, runs on your cloud server.

My implementation was custom, on top of RocksDB. I found it hard to ask LLM to traverse it. While understanding schema of SQLite or making queries to find information is very easy for LLMs. In most cases schema does not have to be inferred since it is going to be available and this makes the job easier. The graph approach may work well for many use-cases but if we want to store structured information for LLMs then SQLite is really good.

I tried a graph based approach in my previous product (1). I am on a new product now and I came back to SQLite. Initially it was because I just wanted a simple DB to enable creating cross-platform desktop apps.

I realized LLMs are really good at using sqlite3 and SQL statements. So in my current product (2) I am planning to keep all project data in SQLite. I am creating a self-hosted AI coding platform and I debated where to keep project state for LLMs. I thought of JSON/NDJSON files (3) but I am gravitating toward SQLite and figuring out the models at the moment (4).

  1. Previous product with a graph data approach https://github.com/pixlie/PixlieAI
  2. Current product with SQLite for its own and other projects data: https://github.com/brainless/nocodo
  3. Github issue on JSON/NDJSON based data for project state for LLMs: https://github.com/brainless/nocodo/issues/114
  4. Github issue on expanding the SQLite approach: https://github.com/brainless/nocodo/issues/141

Still work in progress, but I am heading toward SQLite for LLM state.

This is terrible lack of planning. I understand the need for the US to be strict about people entering the country. They have a right to choose but this is so mismanaged.

There are so many active H-1B visa holders, now everyone is just anxious. The rules can start for new visa applications. For existing holders, there should be a time period where people can figure out if the employer is even able to pay.

If this stays in effect for existing visa holders and the employers cannot pay in time or wants to change the contract, the individuals and their families are stuck. Plus, employees probably lose their job if employer cannot pay the fees.

This may boost remote jobs.

I have worked for many US based startups, all remotely. Timezone difference (I am in India) is a big issue unless the company is very well structured to work asynchronously.

Companies hiring top talent may still hire with a $100K additional charge but even at $250K - 400K salaries, this is a lot of additional cost.

Empowering people is a lovely thing.

Here the author has a passion/side project they have been on for a while. Upgrading the tooling is a great thing. Community may not support this since the niche is too narrow. LLM comes in and helps in the upgrade. This is exactly what we want - software to be custom - for people to solve their unique edge cases.

Yes author is technical but we are lowering the barrier and it will be lowered even more. Semi technical people will be able to solve some simpler edge cases, and so one. More power to everyone.

Wow, didn't know about your product. I gave up city life since the pandemic and have been living a simple life in a small Himalayan village.

I can't overwork anymore for health health. I still lead engineering at an NY based startup remotely in 2022 but work targets made things hard.

I will read your post and apply if it's a fit.

This is the best thing engineers can do. I moved to building as a solo founder. I am building an LLM enabled coding product and I teach. I'm hosting a session on Claude Code today, 134 guests signed up. I'm gradually planning to make money teaching for a few months while building the product.

100%. Like I mentioned in another comment. LLMs should simple close communication and show existing social help options at the first hint of mental distress. This is not a topic where there can be any debate or discussion.

I do not think this is fair. What is fair is at first hint of a mental distress, any LLM should completely cut-off communication. The app should have a button which links to actual help services we have.

Mental health issues are not to be debated. LLMs should be at the highest level of alert, nothing less. Full stop. End of story.