About

First version: Sat Apr 25 2026 19:00:00 GMT-0500 (Central Daylight Time) Last update: Sat Apr 25 2026 19:00:00 GMT-0500 (Central Daylight Time)

About the Owner

If you are looking for an about page on the owner you can find that here.

Origin

This site was a novel idea I had been thinking about for a few days. Around a year ago there was a Dwarkesh Patel podcast I had listened to, I believe on the broad subject of LLMs. My memory of it is quite hazy and really only a specific exchange stuck. That was of Dwarkesh telling the story of his friend who spent hours a day using deep research to increase his own knowledge base. I had considered for quite a while after that how this would fare as opposed to the fast back and forth of regular chat with a model that has access to web search, and I find this to outperform it for several reasons.

Agency

The first of those reasons is agency. I have noticed that the largest negative effect of LLMs is the instant gratification gained from quick positive outcomes. It is a way to get fast gains without core understanding. The nature of this back and forth itself becomes partially thinking into a model and allowing it to finish your thought process for you. You never actually get to form ideas yourself or come to a complete understanding of a topic. Sometimes not even a surface one. Deep Research inherently flips this on its head. It is similar to the relationship of forums to something like Discord. The slower nature of Deep Research requires one to be meditative. To think, to verify, to assess. This is the human part, which is why it strikes me feverishly when people ask questions like "don't I need to check sources?" or "do follow up research myself?" This again is not meant to be a parent, to tell you what to do with no feedback. It is meant to provide high level overviews and point you in directions that are largely verified, and sometimes niche paths that have not been traveled down.

The Meditative Nature of Deep Research

Another huge reason is the largely meditative nature of deep research. It is a slow romantic process of familiarizing yourself with a topic through the exploration of the synopsis and its many sources in real time. It is not just meant to read the compilation itself but the lifeblood it pulled from. In fact quite a few of my earlier deep research prompts were directed at providing dumps of personal blogs that fit a narrow criteria of what I considered to be worthwhile. I found a ton of them useful, and dozens have ended up in my blogroll over time.

Feeding the Knowledge Base

A lot of this was done to feed my notebooks, notes, and LLM wiki which is a new exciting implementation based on Andrej Karpathy's tweet on LLM Knowledge Bases. For those unfamiliar with me, I am a writer, researcher, and mathematics student. With exception to the notes and notebooks above which are modules and extensions to the base, my writings live at krisyotam.com. I am in general a philomath. Lover of learning all things. I have very broad interests and intend to harness that knowledge in the direction of creating interesting essays, blog posts, and curricula. I also have many more ideas for in which direction I could take knowledge work. I am however at the moment continuing to explore the landscape while I deliberate.