Jared Goering

was almost on Sharktank once

[Cernovich @Cernovich]

MAHA was the deciding factor in the Iowa gubernatorial primary. MAHA isn’t a big enough bloc to win, but it can swing an otherwise close race.

[Jared Goering @jaredgoering]

My friend @ZachLahn was elected to be the republican nominee for governor last night!

Super proud to be apart of this awesome team.

[Zach Lahn @ZachLahn]

We did it!!!

Thank you IOWA!

IOWA FIRST!

[Zach Lahn @ZachLahn]

Iowans. Today is the day we shock the establishment across the country.

With support from every Iowan that believes our culture and heritage is worth preserving - we will win.

Even if I have not been your first choice, I’m asking for your support to stop the machine today.

It has been an honor to be in this race and hear from Iowans across the state that we love.

Now is the time.

VOTE IOWA FIRST - Lahn For Governor.

Polls close at 8pm.

[Elijah Bluhm @ElijahBluhm]

19hrs - until IA polls close.

Iowa, VOTE @ZachLahn for Gov. Then…

Call.
Text.
Post.
Drive.
Share.

Get every voter possible.
We can just win this. 🔥🔥

[Steak 'n Shake @SteaknShake]

Steak n Shake endorses Zach Lahn for Governor of Iowa. He is the only MAHA-supporting candidate.

[Dixon Wolfe @dixonswolfe]

Our team @TPAction is proud to endorse @ZachLahn in the crowded field for Iowa Governor.

Zach is a principled MAGA conservative, a sixth generation Iowan, and a businessman who will focus on practical common sense solutions for families and farmers.

@ZachLahn won’t let you down! Iowa, get out and vote on or before Tuesday, June 2nd!

[Jared Goering @jaredgoering]

Charged next to this cool guy today

[Jared Goering @jaredgoering]

I’ve never ran an @openclaw update that didn’t break everything and require a couple of hours to fix.

[Jared Goering @jaredgoering]

Anthropic just showed that from here on, there will be two classes of people. Those who get access to the most powerful models and those who don’t. And they get to decide who’s on the list.

[Anthropic @AnthropicAI]

Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software.

It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans.
https://t.co/NQ7IfEtYk7

[Jared Goering @jaredgoering]

100%. Search is the easy part. Reconciliation is the real memory problem.

Ultramemory is built around atomic facts + typed relations like updates, contradicts, and extends, with provenance on every memory. So when two sources disagree, you don’t just keep “the winner” - you preserve the lineage and can trace why the current fact is current.

Sent a DM. Would love to learn more about what you’re working on.

[Jared Goering @jaredgoering]

@EduWonderer @chamath lol how?

[🕊️ @lichthauch]

Everything around me is rotting, so i build. that is it, that is the whole reason. God built six days and rested one. he could have stopped at function, he could have given us eyes that only see useful things, but he made color. he made sunsets. he made the shape of a woman's back. he made the sound of rain hitting a dirt road at night. beauty was not an afterthought, it was the first thought, everything else came after. i build because the world is falling apart, and a man who does not build is just watching a fire. i lay bricks straight, i put flowers where nobody will see them, i sand wood until the grain is smooth enough to hold without a glove. these are prayers. and the man who says beauty does not matter has never built anything, he has only consumed, and consumption leaves you hollow. everyone i know who only takes has the same eyes, empty, always hungry, looking for the next thing to swallow. hold a hammer instead. hold wood. hold stone. make something that does not need you to survive, and then walk away from it, and feel what that is. that is the closest i have ever got to being alive, and i am not giving it back.

[Jared Goering @jaredgoering]

It's early but it's been genuinely useful for my own research. PRs welcome.

https://t.co/vEeKk3Rhgv

[Jared Goering @jaredgoering]

Saw this and immediately built it. Open-sourced the whole thing:

Ingest URLs, PDFs, tweets, images → LLM compiles a linked markdown wiki → Q&A with citations, knowledge graph, contradiction linting, auto-research, export to HTML/PDF/slides.

Been using it nonstop as a personal research tool. Super useful.

👇

[Andrej Karpathy @karpathy]

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.

[Jared Goering @jaredgoering]

SQLite, runs on localhost. Works with OpenAI, Anthropic, or local models via Ollama. Your data stays on your machine.

pip install lexiconai && lexicon serve

[Jared Goering @jaredgoering]

Under the hood, Lexicon runs on Ultramemory - our multimodal semantic memory engine with embeddings for text, images, audio, and video.

[Jared Goering @jaredgoering]

Set topics to auto-research on a schedule. Lexicon searches the web, ingests new results, recompiles, and lints - your KB stays fresh without you touching it.

[Jared Goering @jaredgoering]

Ask questions against your knowledge base. If the answer isn't confident enough, it automatically searches the web, ingests new sources, and tries again..

[Jared Goering @jaredgoering]

The LLM reads your ingested sources, groups them by topic, and writes interconnected markdown articles with [[wikilinks]] and citations.

New sources get folded into existing articles automatically. You can still edit anything - manual edits are preserved across recompilations.

[Jared Goering @jaredgoering]

Clip any tweet/thread with the extension or paste a URL. Lexicon extracts the text, author, and metadata then compiles it into your wiki alongside everything else.

No more losing that one tweet you bookmarked 6 months ago.