5 ms·
This is fantastic work. The focus on a local, sandboxed execution layer is a huge piece of the puzzle for a private AI workspace. The `coderunner` tool looks in
by andylizf 1y ago
This is fantastic work. The focus on a local, sandboxed execution layer is a huge piece of the puzzle for a private AI workspace. The `coderunner` tool looks incredibly useful.
A complementary challenge is the knowledge layer: making the AI aware of your personal data (emails, notes, files) via RAG. As soon as you try this on a large scale, storage becomes a massive bottleneck. A vector database for years of emails can easily exceed 50GB.
(Full disclosure: I'm part of the team at Berkeley that tackled this). We built LEANN, a vector index that cuts storage by ~97% by not storing the embeddings at all. It makes indexing your entire digital life locally actually feasible.
Combining a local execution engine like this with a hyper-efficient knowledge index like LEANN feels like the real path to a true "local Jarvis."
Code: https://github.com/yichuan-w/LEANN https://github.com/yichuan-w/LEANN
Paper: https://arxiv.org/abs/2405.08051 https://arxiv.org/abs/2405.08051
- sebmellen 1y agoI know next to nothing about embeddings. Are there projects that implement this same “pruned graph” approach for cloud embeddings?
- NJL3000 1y agoIt’s in the works… been meaning to do a show HN moment to see if it flies or I Fall on my face..
- doctoboggan 1y ago> A vector database for years of emails can easily exceed 50GB. In 2025 I would consider this a relatively meager requirement.
- andylizf 1y agoYeah, that's a fair point at first glance. 50GB might not sound like a huge burden for a modern SSD. However, the 50GB figure was just a starting point for emails. A true "local Jarvis," would need to index everything: all your code repositories, documents, notes, and chat histories. That raw data can easily be hundreds of gigabytes. For a 200GB text corpus, a traditional vector index can swell to >500GB. At that point, it's no longer a "meager" requirement. It becomes a heavy "tax" on your primary drive, which is often non-upgradable on modern laptops. The goal for practical local AI shouldn't just be that it's possible, but that it's also lightweight and sustainable. That's the problem we focused on: making a comprehensive local knowledge base feasible without forcing users to dedicate half their SSD to a single index.
- notsylver 1y agoYou already need very high end hardware to run useful local LLMs, I don't know if a 200gb vector database will be the dealbreaker in that scenario. But I wonder how small you could get it with compression and quantization on top
- varenc 1y ago> You already need very high end hardware to run useful local LLMs A basic macbook can run gpt-oss-20b and it's quite useful for many tasks. And fast. Of course Macs have a huge advantage for local LLMs inference due to their shared memory architecture.
- OneDeuxTriSeiGo 1y agoYou can already do A LOT with an SLM running on commodity consumer hardware. Also it's important to consider that the bigger an embedding is, the more bandwidth you need to use it at any reasonable speed. And while storage may be "cheap", memory bandwidth absolutely is not.
- wafflemaker 1y agoI'm no dev either and still set up remote ssh login to be able to use LaTeX at home PC from my laptop. Also, with many games and dual boot on my gaming PC I still have some space left on my 2TB NVME SSD. And my not enthusiast MOBO could fit two more. It took so much time to install LaTeX and packages, and also so much space, my 128GB drive couldn't handle it.
- mwcz 1y agoI've worked in other domains my whole career, so I was astonished this week when we put a million 768-len embeddings into a vector db and it was only a few GB. Napkin math said ~25 GB and intuition said a long list of widely distributed floats would be fairly uncompressable. HNSW is pretty cool.
- derefr 1y agoThe mid-spec 2025 iPhone can run “useful local LLMs” yet has 256GB of total storage. (Sure, this is a spec distortion due to Apple’s market-segmentation tactics, but due to the sheer install-base, it’s still a configuration you might want to take into consideration when talking about the potential deployment-targets for this sort of local-first tech.)
- snoman 1y agoTake whatever you're indexing and make it 16-20x and that’s a good approximation of what the vector db’s total size is going to be.
- jononor 1y agoWhy is it like that, currently? There is no information added by a vector index compared to the original text. And the text is highly redundant and compressible with even lossless functions. Furthermore a vector index is already lossy and approximate. So conceptually it is at least possible to have an index that would be a fraction of the size of what is indexed?
- snoman 1y agoThere is some information added, depending on the vector db and context (some systems will add permissions related metadata so that the LLM won’t pull chunks that the user didn’t have access to). The vector itself is pretty large (512 dimensions). The chunks have an overlap (iirc 30% but someone feel free to correct me). I don’t _think_ the data is typically compressed (not sure why but I assume performance).
- oblio 1y agoIt feels weird that the search index is bigger than the underlying data, weren't search indexes supposed to be efficient formats giving fast access to the underlying data?
- andylizf 1y agoExactly. That's because instead of just mapping keywords, vector search stores the rich meaning of the text as massive data structures, and LEANN is our solution to that paradoxical inefficiency.
- yichuan 1y agoI guess for semantic search(rather than keyword search), the index is larger than the text because we need to embed them into a huge semantic space, which make sense to me
- iezepov 1y agoGood point! Maybe indexing is a bad term here, and it's more like feature extraction (and since embeddings are high dimensional we extract a lot of features). From that point of view it makes sense that "the index" takes more space than the original data.
- catlifeonmars 1y agoWhy would the embeddings be higher dimensionally than the data? I imagine the embeddings would contain relatively higher entropy (and thus lower redundancy) than many types of source data.
- cm228 1y agodepends on the chunk-size used to create the embedding.
- brookst 1y agoNonclustered indexes in RDBMS can be larger than the tables. It’s usually poor design or indexing a very simple schema in a non-trivial way, but the ultimate goal of the index is speed, not size. As long as you can select and use only a subset of the index based on its ordering it’s still a win.
- wfn 1y agoThank you for the pointer to LEANN! I've been experimenting with RAGs and missed this one. I am particularly excited about using RAG as the knowledge layer for LLM agents/pipelines/execution engines to make it feasible for LLMs to work with large codebases. It seems like the current solution is already worth a try. It really makes it easier that your RAG solution already has Claude Code integration![1] Has anyone tried the above challenge (RAG + some LLM for working with large codebases)? I'm very curious how it goes (thinking it may require some careful system-prompting to push agent to make heavy use of RAG index/graph/KB, but that is fine). I think I'll give it a try later (using cloud frontier model for LLM though, for now...) [1]: https://github.com/yichuan-w/LEANN/blob/main/packages/leann-mcp/README.md https://github.com/yichuan-w/LEANN/blob/main/packages/leann-...
- mccoyb 1y agoThat can't be the correct paper... I think you meant this: https://arxiv.org/abs/2506.08276 https://arxiv.org/abs/2506.08276
- andylizf 1y agoYeah that's it. My bad lol
- johnfn 1y agoNo no, getting your entire workflow local requires solving P=NP.
- antoniojtorres 1y agoWait how?
- janderson215 1y agoJust some sarcasm. You can safely disregard if you didn’t get a chuckle.
- unixhero 1y agoI have 26tb hardrives, 50gb doesnt scare me. Or should I be?
- technocratius 1y agoI think you'd want things in RAM for performance reasons but would love to be corrected by people with more knowledge/experience on the subject
- unixhero 1y agoOh the number was memory space? That changes the maths a little bit. But I do have 50gb available for a model no problem whatsoever. 384gb is the new 32gb.
- wy1346 1y agoThis looks incredibly useful for making large-scale local AI truly practical.
- jychang 1y agoThis is annoyingly Apple-only though. Even though my main dev machine is a Macbook, this would be a LOT more useful if it was a Docker container. I'd still take a Docker container over an Apple container, because even though docker is not VM-level-secure, it's good enough for running local AI generated code. You don't need DEFCON Las Vegas levels of security for that. And also because Docker runs on my windows gaming machine with a fast GPU with WSL ubuntu, and my linux VPS in the cloud running my website, etc etc. And most people have already memorized all the basic Docker commands. This would be a LOT better if it was just a single docker command we can copy paste, run it a few times, and then delete if necessary.
- glhaynes 1y agoI’m no expert on these things, but since Apple Containerization uses OCI images, I’d think you’d be able to sub in Docker (or Podman, etc) as the runtime pretty trivially. Like Podman, it uses a very similar command line interface to Docker’s. Edit: Oh, I see now that Coderunner is Apple Containerization-specific.
- psychoslave 1y agoWhy is that considred relevant to get a RAG of people digital traces burdening them in every single interactions they have with a computer? Having locally distributed similar grounds is one thing. Push everyone to much in its own information bubble, is an other orthogonal topic. When someone mind recall about that email from years before, having the option to find it again in a few instants can interesting. But when the device is starting to funnel you through past traces, then it doesn't matter much whether it the solution is in local or remote: the spontaneous thought flow is hijacked. In mindset dystopia, the device prompts you.
- solarkraft 1y agoSince it’ll be local, this behavior can be controlled. I for one find the option of it digging through my personal files to give me valuable personal information attractive.
- OldfieldFund 1y agoI'm gonna put it here for visibility: Use patchright instead of Playwright: https://github.com/Kaliiiiiiiiii-Vinyzu/patchright https://github.com/Kaliiiiiiiiii-Vinyzu/patchright
- bamboozled 1y agoWhat problem does patchright solve?
- jdelsman 1y agoNot being detected by things like bot detection.