BrandiATMuhkuh

haha how cool.

Not sure if I’m out of touch (not doing robotics right now) or if things just slowed down. Back in ~2013 it looked like humanoid robots were just around the corner. Boston Dynamics became a hit. Aldebaran’s Pepper was presented. And the DARPA Robotics Challenge was on everyone’s lips.

And then suddenly, nothing.

That is so awesome. I wish I had that 11 years ago during my Microsoft Research Internship. I had a NAO robot (from another great French company) and it walked to employees to motivate them to do some movements. To prevent long term injuries from being in the same position for too long. It had a visual compass but it didn't work. And just telling to "walk straight" clearly didn't work :D Here is the video in case someone is curious: https://www.youtube.com/watch?v=aZkmNqpJCcM&t=190s (and the paper: https://brandstetter.io/papers/Fidgebot_Working_Out_while_Wo...)

What a simple and smart idea. Wonderful

Pretty cool. I always thought merges should happen by comparing the AST and not lines

I've been using Superset as my daily driver for about 2 months and I'm loving it.

i have created a simple setup script that reserves a couple of ports per workspace/worktree. with that, each ticket/workspace can starts it's own webserver. Allowing me to work on multiple things in parallel. that also allows me to use Claude+Chrome MCP to e2e test each implementation in parallel.

Besides my own repo I also have a couple open source repos open. so I can quickly ask Claude questions about it instead of looking at the docs, which are usually outdated.

All in all, I usually have ~5 workspace/tickets/agents running at the same time. And one as Q&A agents per OSS repo

That's epic. Will try it ASAP.

Was look for exactly that (but for Features). currently I'm using superset.sh which works great. But the problem is that such tools need to re-implement everything (browser, terminal, etc.) while a "VM-like" approach doesn't.

It is possible to "clone" a drawer that auto starts server, browser, etc. So I can start new PRs quickly and jump between them while the agents run in the background?

since the end of last year we have had instant payment in the EU. no more need for Mastercard/VISA. what I see missing is education.

That's why I'm telling each merchant I come bye to use it. super simple.

we tried it just before. it's interesting what it does. writing lots of python scripts.

however the result (excel/spreadsheet) looks different each time you run it. Which is annoying when you run it at the end of each month.

btw: this is not surprising when you look at the low details the skills have.

Cool project. Just yesterday I looked at https://portless.sh/ which is vercels take.

Recently I've started to use https://superset.sh as alternative to Warp. After the volks @mastra mentioned it. Very cool open source project.

I'm actually pretty proud of the final setup I've created with it.

Each time I start to implement a new ticket, superset will pull the ticket from linear, create a worktree/workspace, reserve ports, start the servers, start a browser and start Claude with the ticket as instructions.

The cool thing with this setup is, I can have like 10x the same servers running on different ports/worktrees. Each time an agent is done, I switch to the workspace, look at the browser and can immediately test things.

It's like having 10 virtual desktops. Wonderful!

Very cool. And congrats on the launch.

I started to use superset 2 days ago. Which seems similar. It's pretty nice: https://superset.sh

Fyi: here are some things I would like to have for such a tool - notification when an agent is done - each tabs/space has its own terminal, browser, agent - each tab/space runs in a sandbox (eg docker) - each tab/space can run my dev server. But must not conflict with the other dev servers running - each tab/space has a mcp server for the built in browser

Nice to have: - remote access against my machine/tabs - being able to make screenshots

I'm curious how you manage https. If OneCLI intercepts all traffic from the agent (harness/tools/...) and then replaces parts with other data, it should break https.

Or is that a man in the middle "attack". And users have to install a certificate?

Very cool. Was looking for something like this for a new project of mine. (I'm working on a project that is like a marriage of retool+OpenClaw. It's used by SME to quickly build inhouse apps)

Pretty cool work! Question: what's the difference in using BUN? I'm currently using buns react frontend/backend system. And afaik it's also written in rust.

The part that I don't see are the 'use server' .

   Location: Linz/Austria/EU 
   Remote: OK (Any timezone OK)
   Willing to relocate: NO 
   Technologies: AI (LLM, RAG, Agents, Voice, AI-SDK, Mastra), Typescript, React, Next.js, Supabase, Firebase, Postgres, GCP
   Résumé: https://brandstetter.io/Resume_Brandstetter_Jurgen.pdf
   Email: j@brandstetter.io
   LinkedIn: https://www.linkedin.com/in/j-brandstetter
   Salary: USD 150k / year

I'm very product/customer-focused and pragmatic. I like to move super fast (not only because of Cursor :D).

My dream environment is an early-stage, fully remote startup.

PS: I have a family, which means the main compensation via shares isn't an option for me.

https://brandstetter.io/ Super outdated (last update ~ 10y ago). Still counts

Congratulations on the launch.

One customer type that will absolutely love that are architecture studios. Basically all data they generate (drawing, plans, presentation) live either on SharePoint or NFS (or some other file system like ACC).

If you can provide for them a solution they can host on their private cloud (air-gapped) you have an enterprise-deal.

Why air-gapped - many documents they generate are secret (military or competition. Think next largest tower in the world) - copy right: for example the German norm body DIN does not allow you to use documents from them and add them into an LLM (I'm not talking training, I simply talk RAG) except!!! if you keep it in a private cloud like Azure.

Why do I know that: because I worked on that problem @howie.systems

I'm using this approach quite often. I don't know of any documents created by humans for humans that have no formatting. The formatting, position etc. are usually an important part of the document.

Since the first multimodal llms came out, I'm using this approach when I deal with documents. It makes the code much simpler because everything is an image and it's surprisingly robust.

Works also for embeddings (cohere embed v4)

Congratulations on the launch.

I was recently trying to tackle the same problem (@howie.systems). The hardest 2 problems we had to face were ACL and large files (and large volumes).

How did you solve the ACL part? I worked with a customer that had 200k pdf/images/dwg files on SharePoint and other 1M on samba. It took like a week to sync it all and keep tabs on the access rights of each employee.

How did you solve unpredictable large files: a pdf 2000pages, maybe some A0 in the mix. Or some 4GB power point presentations?

PS: great fan of gather. PPS: say hi to Clinton from me (amy.app) if he is still around. He was our mentor back in New Zealand at the flux accelerator (2016)

Congratulations on the launch. I think it's a smart move to not use MCP here. Because your LLM really needs to understand how the different integrations work together.

Question: you say you do semantics search. If I understand correctly that means you must somehow index all data (Gmail, GDrive, ...) otherwise the AI would have to "download/scan" thousands of files each time you ask a question. So how do you do the indexing?

For some background: I'm working on something similar. My clients are architects. They have about 300k files for just one building. With an added 50k issues and a couple of thousand emails. And don't forget all subcontractors.

Would Slashy be able to handle that?

Howie.systems (EU VC funded) | Full-Stack (AI) Engineer | ONSITE Vienna, Austria | Full-time | Start date: ASAP

- We’re building Howie.systems, an AI platform for the architecture/engineering/construction industry. Our focus: automating knowledge extraction and retrieval from large document sets (100k+ files, multi-tenant, multi-user, with Supabase/Postgres + pgvector under the hood). We’re fully funded by international investors and are expanding our team in Vienna.

Role: We’re looking for a Full-Stack Engineer with a love for strong typing and AI.

- Must: Next.js, Supabase, very TypeScript-safe mindset

- Super plus: experience with AI frameworks (Vercel AI SDK, Mastra, LangChain, etc.)

- Should have good experience with AI Coding tools like Cursor, Cloud Code

You’ll help us scale our ingestion, RAG, and AI interaction layers into production-grade tools dealing with enterprise customers

What we offer:

- Competitive salary + equity possible Onsite in Vienna (no remote option)

- Small, hands-on team with big international backing Work on a fully modern stack (TS, Supabase, Vercel, AI frameworks)

If this sounds like you, send your CV and a cover letter to: contact@howie.systems Please include within the cover letter, detailing your experience and why you think you would be a fit for the role

Does this also work if I have data on SharePoint, DropBox, etc. and want to pull them (sync with local machine)?

My use case is mostly ETL related, where I want to pull all customers data (enterprise customer) so I can process them. But also keep the data updated, hence pull?

I'll for sure take a deeper look. Ingestion has been by far the biggest pain and least fun. Those infra parts hold us back from the cool things -> building agents/search

Chroma looks cool. Congratulations on the Cloud version.

For my client I've "built" a similar setup with Supabase + pgVector and I give the AI direct SQL access.

Here is the hard part: Just last week did I index 1.2 million documents for one project of one customer. They have pdfs with 1600 pages or PPTX files of >4GB. Plus lots of 3D/2D architecture drawings in proprietary formats.

The difficulty I see is - getting the data in ETL. This takes days and is fragile - keep RBAC - Supabase/pgVector needs lots of resources when adding new rows to the index -> wish the resources scale up/down automatically. Instead of having to monitor and switch to the next plan.

How could chroma help me here?

I have started treading everything as images when multimodal LLMs appeared. Even emails. It's so much more robust. Especially emails are often used as a container to send a PDF (e.g. a contract) that contains an image of a contract that was printed. Very very common.

I have just moved my company's RAG indexing to images and multimodal embedding. Works pretty well.

   Location: Linz/Austria/EU 
   Remote: OK (Any timezone OK)
   Willing to relocate: NO 
   Technologies: AI (LLM, RAG, Agents, Voice, AI-SDK), Typescript, React, Next.js, Supabase, Firebase, Postgres, GCP
   Résumé: https://brandstetter.io/Resume_Brandstetter_Jurgen.pdf
   Email: j@brandstetter.io
   LinkedIn: https://www.linkedin.com/in/j-brandstetter
   Salary: USD 150k / year

I'm very product/customer-focused and pragmatic. I like to move super fast (not only because of Cursor :D).

My dream environment is an early-stage, fully remote startup.

PS: I have a family, which means the main compensation via shares isn't an option for me.

Those things are not mutually exclusive. We use RAG and Vector stores to index terabyte of data. Then use tools calls (MCP) to allow the AI to write SQL to directly query the data (vector store).

I was just building a SharePoint integration for some enterprise customers (I do RAG on their data) and I find it brutal, that now, I have access to all their SharePoint data for all SharePoint sites. Even the ones I don't want to index. And I even use user login over admin/service key login.

AFAIK, the Oauth claims of SharePoint don't allow specifying particular projects only. (BTW: same counts for platforms like ACC/BIM360)

Congratulations on the launch! This is really cool.

I'm not super active in the humanoid robot space anymore, however I did my PhD about 9 years ago in HRI. That was the time of Boston Dynamics, DARPA robotics challenge, and Aldebaran's Pepper and Nao robots.

You mentioned you are building everything open source. What happened with ROS and related projects? Do you build on top of that, or is that all super outdated that a reboot was needed?

Another question I have is: why are you choosing a two-legged human over a four-legged one?

My experiments with two legged robots were mostly bad. Not only did they fall basically all the time but they also had a big drift. So far, I have not seen any large improvements. But again, I might be very outdated.

I always said to my colleagues. The main point stopping robots from picking up is a stable platform. And with the platform I mean walking.

   Location: Linz/Austria/EU 
   Remote: OK (Any timezone OK)
   Willing to relocate: NO 
   Technologies: AI (LLM, RAG, Agents, Voice, AI-SDK), Typescript, React, Next.js, Supabase, Firebase, Postgres, GCP
   Résumé: https://brandstetter.io/Resume_Brandstetter_Jurgen.pdf
   Email: j@brandstetter.io
   LinkedIn: https://www.linkedin.com/in/j-brandstetter
   Salary: USD 150k / year

Currently, I'm designing and working on an agentic enterprise search engine for the AEC industry.

I'm very product/customer-focused and pragmatic. I like to move super fast (not only because of Cursor :D).

My dream environment is an early-stage, fully remote startup.

PS: I have a family, which means the main compensation via shares isn't an option for me.