Ayan Khanna

18 | NYC | prev @joinsaturn (sold to @snap)

[Ayan Khanna @ayan_khanna21]

@henrylangmack “Soar points” seems worth exploring

[Ayan Khanna @ayan_khanna21]

Real estate agents are (90% of the time) useless.

Having to apply separately to apartments rather than having a centrally verified single application is stupid.

The same centralized check that you’re not a bum should let you self-tour.

^ notes from nyc, someone build

[Ayan Khanna @ayan_khanna21]

@blakeandersonw Im pretty bought into post-AGI infinite abundance theory, and that the status game will shift primary variable from money to health/longevity. Obviously social pull and general “power” over people will also stay

[Ayan Khanna @ayan_khanna21]

Yep ever since I got my whoop everytime I pick up a drink I pause to think about how it’s going to screw up my recovery the next day.

Then I pound it anyways, but worth noting.

[Murray Hill Guy @MurrayHillGuy1]

Noticing a massive shift out in NYC lately… people just aren’t drinking like they used to.

And if they are, it’s like one drink they nurse for 3 hours instead of full blackout mode.

Feel like a lot of people late 20’s/30’s have hit that point where waking up with crippling anxiety, being useless the entire next day, spending $200+ at bars and feeling inflamed for 48 hours just stopped feeling worth it.

Obviously people still want to go out, flirt, see friends, meet people, but they also want to wake up early, get a workout in, go on a run, look good and not feel like death.

Also think the explosion of fitness culture, run clubs, wellness trends, GLP-1s, whoops, sleep tracking, etc changed an entire generation’s mindset around alcohol.

Few years ago getting blacked out every weekend was almost a flex and the norm. Now people care more about being lean, productive and mentally sharp.

Honestly feels like NYC nightlife culture is changing in real time.

[Ayan Khanna @ayan_khanna21]

Unfortunately my $FIG likes to ignore the second half

[Ivan Boroja @ivanboroja]

Here we go again

[Ayan Khanna @ayan_khanna21]

IMO user knowledge is still an unsolved problem for LLMs

Your best friend can tell me exactly how you’d react to something. Why can’t ai?

Memories ≠ Personality Graph

RAG injects events, but useless w/o large quantity and time to reason

I want a literal second brain

[Ayan Khanna @ayan_khanna21]

How do you combine this high-level, quick indexing of thoughts with NotebookLM level context (the fancy RAG he mentions) to create a real second brain i can engage in extended conversation with

[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.

[Ayan Khanna @ayan_khanna21]

@willahmed can we please get an official MCP server for Whoop

[Ayan Khanna @ayan_khanna21]

Opening a new savings account for Mythos tokens

[Ayan Khanna @ayan_khanna21]

How long before ai turns OF into a male dominated industry

[Ayan Khanna @ayan_khanna21]

Who knows someone that's crazy knowledgable about business tax structuring. I need 15 minutes, Claude seems somewhat stupid on this subject

[Ayan Khanna @ayan_khanna21]

I learn more everyday just how much the average consumer hates AI

[Ayan Khanna @ayan_khanna21]

@trq212 "Claude make a random commit every few hours so my github looks cracked"

[Ayan Khanna @ayan_khanna21]

I've heard a lot of people say small businesses will slowly aggregate into either AI labs or huge enterprises.

I'd bet on the opposite - uptick of SMBs as traditional industries outside tech realize a small team can punch above their weight class

[Ayan Khanna @ayan_khanna21]

This isn't saying its bad (very helpful for non-devs), but just take the time to export to an IDE and touch it up

[Ayan Khanna @ayan_khanna21]

Please stop shipping websites directly from Lovable they look shit

[Ayan Khanna @ayan_khanna21]

@aslater @zach_yadegari Me and @hugostegletz had this exact same debate a few months ago I’m team slater on this one

[MLFootball @MLFootball]

THIS SLO-MOTION VIDEO CLEARLY SHOWS THAT BRANDIN COOKS CAUGHT THE FOOTBALL AND LANDED ON THE GROUND.

THE #BILLS GOT SCREWED.

[Ayan Khanna @ayan_khanna21]

@Batmantallk Girlfriend

[Ayan Khanna @ayan_khanna21]

@mntruell Intelligent switches between ask/agent. Big point to antigravity for this.

On cursor if I ask a clarifying question in agent mode it almost always finds an excuse to edit some code