LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:
Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.
IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).
Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.
Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.
Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.
Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.
Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.
TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.
Conversation
![Amitav Krishna](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="48" height="48"><rect fill-opacity="0"/></svg>)
Post your reply
I did even more.
Ingested everything on X about AI and had it build this: alignednews.com/ai
Has a feed for your AI to hit too.
Includes papers, models, news, events, and much more.
Updated three times a day. Reads EVERYONE in AI on X and 8,300 AI companies too.
i made a makrdown package to help ppl become poweruser of obsidian x ai where i operationalized some of what you describe: superpaper.ai
there's a also a skill for superpaper you can install using `npx skills add superinterface-labs/superpaper`
Can you make a video of how you work with md files, agentic IDEs?
Your earlier explanations of using LLMs were really helpful.
Can you share more on the incremental compilation?
I've found that if processing one by one, they don't have enough context to understand how to divide to directories.
Is there an optimal batch size? Multiple stages?
Atm it's not a fully autonomous process, I add every source manually, one by one and I am in the loop, especially in early stages. After a while, the LLMs "gets" the pattern and the marginal document is a lot easier, I just say "file this new doc to our wiki: (path)".
Every business has a raw/ directory.
Nobody's ever compiled it.
That's the product.
Currently no because I'm trying to keep it super simple and flat, it's just a nested directory of .md files and .png files and a few .csv and .py, and the schema is kept up to date in AGENTS.md . The LLMs get this very easily. Any custom functions are easy to vibe code tools for.
Same, I have a similar setup. A mix of Obsidian, Cursor (for md), and vibe-coded web terminals as front-end.
Since I do a podcast, the number/diversity of research interests is very large. But the knowledge-base approach has been working great.
For answers, I often have it
Redefine your wellness without touching your bank account.
Did you know you can use your HSA/FSA pre-tax dollars for your CUDIS ring? Level up your recovery and anti-aging routines before your funds expire!
0:19
CUDIS is HSA/FSA Eligible!
I like this approach because it mitigates the contamination risks of agent-generated content in your primary vault... the agents need a playground too!
Quote
kepano
@kepano
I like @karpathy's Obsidian setup as a way to mitigate contamination risks. Keep your personal vault clean and create a messy vault for your agents.
I prefer my personal Obsidian vault to be high signal:noise, and for all the content to have known origins.
Keeping a separation x.com/karpathy/statu…
I have also been obsessed with building LLM knowledge bases.
Here is one example of the type of things you can do that Karpathy is alluding to: x.com/omarsar0/statu
LLMs are excellent at curating and searching (finding connections) once data is stored properly.
Quote
elvis
@omarsar0
Been exploring a new way to explore AI research papers to discover deeper insights.
Agents are at the center of it.
So far, I've built this little interactive artifact generator in my orchestrator to visualize things.
This allows me to change views and insights (on-demand)
turns out my strategy of not maintaining notes for decades was just ahead of its time. LLMs will do it for me now.
Yes, totally.
I have the LEARN phase of the PAI algorithm determine if something should be a Knowledge article, and it goes and constructs it for us under MEMORY/KNOWLEDGE.
So anything we've been working on in a session gets harvested as a Knowledge article, and it's always on
I've been on this exact setup for about a year now - the biggest unlock imo is you can synthesize any k number of topics across any k domains and the possibility becomes O(n^k)
For a 500-note vault:
k=2: 250,000 ordered pairs
k=3: 125 million paths
k=4: 62.5 billion
you can
just made a tldr with arch visualization:
Quote
himanshu
@himanshustwts
and here is the full architecture of the LLM Knowledge Base system covering every stage from ingest to future explorations. x.com/karpathy/statu…
Exactly! No greater alpha than using LLMs to refine one's conceptual models!
Quote
Carlos E. Perez
@IntuitMachine
I've finally resolved how Quaternion Process Theory of Cognition is connected with Alexander's Nature of Order.
There’s a new version of this post
This month i spent a week writing scripts to organize my files with local LMs (Qwen 4B)
Lots of image/video tagging, captions, labels, etc. wrapped around a cli tool to easily context search.
Very fun. Now my coding agent can search files better and help me content write.
Yep! I've been constructing knowledge graphs a while using notations like RDF-Turtle, JSON-LD, and Markdown.
A recent example using , which showcases the power of LLMs, in this regard.
x.com/i/grok/share/3
