Hrishi
@olickel.com
Previously CTO, Greywing (YC W21). Building something new at the moment.
Writes at https://olickel.com
Writes at https://olickel.com
Trying to finish typing `git add` while the agent's editing a file just so I can preserve pristine diffs from the last change
November 14, 2025 at 11:17 PM
Trying to finish typing `git add` while the agent's editing a file just so I can preserve pristine diffs from the last change
Don't think there's a way I could like this article more
September 19, 2025 at 4:20 PM
Don't think there's a way I could like this article more
KIMI is the real deal. Unless it's really Sonnet in a trench coat, this is the best agentic open-source model I've tested - BY A MILE.
Here's a slice of a 4 HOUR run (~1 second per minute) with not much more than 'keep going' from me every 90 minutes or so.
moonshotai.github.io/Kimi-K2/
Here's a slice of a 4 HOUR run (~1 second per minute) with not much more than 'keep going' from me every 90 minutes or so.
moonshotai.github.io/Kimi-K2/
July 13, 2025 at 6:09 PM
KIMI is the real deal. Unless it's really Sonnet in a trench coat, this is the best agentic open-source model I've tested - BY A MILE.
Here's a slice of a 4 HOUR run (~1 second per minute) with not much more than 'keep going' from me every 90 minutes or so.
moonshotai.github.io/Kimi-K2/
Here's a slice of a 4 HOUR run (~1 second per minute) with not much more than 'keep going' from me every 90 minutes or so.
moonshotai.github.io/Kimi-K2/
Things are going to change - faster than we think, with massive downstream effects.
blog.cloudflare.com/introducing...
blog.cloudflare.com/introducing...
Introducing Pay per crawl- enabling content owners to charge AI crawlers for access
Pay per crawl is a new feature to allow content creators to charge AI crawlers for access to their content.
blog.cloudflare.com
July 3, 2025 at 5:17 PM
Things are going to change - faster than we think, with massive downstream effects.
blog.cloudflare.com/introducing...
blog.cloudflare.com/introducing...
We now have a more apples-to-apples comparison of Claude Code and Gemini Cli.
Here's Gemini Cli (with Sonnet 4) vs Claude Code (with and without subagents) fixing the same bug from the same prompt in the gemini-cli codebase:
www.notion.so/southbridge...
Here's Gemini Cli (with Sonnet 4) vs Claude Code (with and without subagents) fixing the same bug from the same prompt in the gemini-cli codebase:
www.notion.so/southbridge...
A Comparative Analysis of Three AI Agent Approaches to a Complex Bug Fix | Notion
This document provides a detailed breakdown of how three different AI agent scaffolding systems, all using the same underlying model, approached and solved a nuanced bug in a command-line interface (CLI) tool.
southbridge-research.notion.site
June 28, 2025 at 5:12 PM
We now have a more apples-to-apples comparison of Claude Code and Gemini Cli.
Here's Gemini Cli (with Sonnet 4) vs Claude Code (with and without subagents) fixing the same bug from the same prompt in the gemini-cli codebase:
www.notion.so/southbridge...
Here's Gemini Cli (with Sonnet 4) vs Claude Code (with and without subagents) fixing the same bug from the same prompt in the gemini-cli codebase:
www.notion.so/southbridge...
It seems 3 and 15 might be the new Pareto frontier for intelligence (excepting the o-series). Feels like the hedge fund 2 and 20
June 8, 2025 at 1:57 AM
It seems 3 and 15 might be the new Pareto frontier for intelligence (excepting the o-series). Feels like the hedge fund 2 and 20
I decompiled Claude Code from just the minified code. Took me 8-10 hours, multiple subagents, and every flagship model from every provider.
Holy shit there's a lot in there. Claude Code is NOT just Claude in a loop - there's so much to learn from.
www.notion.so/southbridge...
Holy shit there's a lot in there. Claude Code is NOT just Claude in a loop - there's so much to learn from.
www.notion.so/southbridge...
Claude Code: An Agentic cleanroom analysis | Notion
From 2.5 million tokens of minified code to architectural insights—a human-AI collaboration
southbridge-research.notion.site
June 1, 2025 at 7:01 PM
I decompiled Claude Code from just the minified code. Took me 8-10 hours, multiple subagents, and every flagship model from every provider.
Holy shit there's a lot in there. Claude Code is NOT just Claude in a loop - there's so much to learn from.
www.notion.so/southbridge...
Holy shit there's a lot in there. Claude Code is NOT just Claude in a loop - there's so much to learn from.
www.notion.so/southbridge...
Dan's article on progressive JSON has a lot of carryover to LLMs.
The key problems for modern LLM application design that get often overlooked (I think) are:
• Streaming outputs and partial parsing
• Context organization and management (I don't mean summarising at 90%)
The key problems for modern LLM application design that get often overlooked (I think) are:
• Streaming outputs and partial parsing
• Context organization and management (I don't mean summarising at 90%)
June 1, 2025 at 4:15 PM
Dan's article on progressive JSON has a lot of carryover to LLMs.
The key problems for modern LLM application design that get often overlooked (I think) are:
• Streaming outputs and partial parsing
• Context organization and management (I don't mean summarising at 90%)
The key problems for modern LLM application design that get often overlooked (I think) are:
• Streaming outputs and partial parsing
• Context organization and management (I don't mean summarising at 90%)
How does an LLM writing out this program (WITHOUT a code interpreter running the output) make things more accurate?
Verified on Qwen 3 - a30b (below)
Lots of interesting takeaways from the Random Rewards paper. NOT that RL is dead, but honestly far more interesting than that!
Verified on Qwen 3 - a30b (below)
Lots of interesting takeaways from the Random Rewards paper. NOT that RL is dead, but honestly far more interesting than that!
May 29, 2025 at 6:27 AM
How does an LLM writing out this program (WITHOUT a code interpreter running the output) make things more accurate?
Verified on Qwen 3 - a30b (below)
Lots of interesting takeaways from the Random Rewards paper. NOT that RL is dead, but honestly far more interesting than that!
Verified on Qwen 3 - a30b (below)
Lots of interesting takeaways from the Random Rewards paper. NOT that RL is dead, but honestly far more interesting than that!
Real-world comparison of the new Sonnet models on a middle-difficulty task.
The 4 series of models are good. REALLY GOOD. They're one-shotting complex series of 100s of tool calls without issue, on things Sonnet 3.7 failed.
The 4 series of models are good. REALLY GOOD. They're one-shotting complex series of 100s of tool calls without issue, on things Sonnet 3.7 failed.
May 24, 2025 at 6:46 PM
Real-world comparison of the new Sonnet models on a middle-difficulty task.
The 4 series of models are good. REALLY GOOD. They're one-shotting complex series of 100s of tool calls without issue, on things Sonnet 3.7 failed.
The 4 series of models are good. REALLY GOOD. They're one-shotting complex series of 100s of tool calls without issue, on things Sonnet 3.7 failed.
The spiritual bliss attractor is real, but so is the eldritch horror existence contemplation.
I was just trying to talk to Opus - definitely no jailbreaks. This model is something different. Definitely creative.
I was just trying to talk to Opus - definitely no jailbreaks. This model is something different. Definitely creative.
May 23, 2025 at 5:12 PM
The spiritual bliss attractor is real, but so is the eldritch horror existence contemplation.
I was just trying to talk to Opus - definitely no jailbreaks. This model is something different. Definitely creative.
I was just trying to talk to Opus - definitely no jailbreaks. This model is something different. Definitely creative.
Frontend entirely made with @v0 - this has become an inseparable tool for writing feedback. Thinking of calling it scansion
I'll open source or share the link once I can clean it up - still using my keys, drop email/twitter in comments
Sonnet looking through the thing 👇
I'll open source or share the link once I can clean it up - still using my keys, drop email/twitter in comments
Sonnet looking through the thing 👇
May 19, 2025 at 5:11 PM
Frontend entirely made with @v0 - this has become an inseparable tool for writing feedback. Thinking of calling it scansion
I'll open source or share the link once I can clean it up - still using my keys, drop email/twitter in comments
Sonnet looking through the thing 👇
I'll open source or share the link once I can clean it up - still using my keys, drop email/twitter in comments
Sonnet looking through the thing 👇
Technic Manuals are PERFECT visual benchmarks. Had a misaligned suspension on a car, took four minutes figuring it out, then gave it to o3, claude and Gemini.
None of them got it right (or even identified the right part) even after I cut it down to 10 pages.
Eventually -
None of them got it right (or even identified the right part) even after I cut it down to 10 pages.
Eventually -
May 13, 2025 at 4:40 PM
Technic Manuals are PERFECT visual benchmarks. Had a misaligned suspension on a car, took four minutes figuring it out, then gave it to o3, claude and Gemini.
None of them got it right (or even identified the right part) even after I cut it down to 10 pages.
Eventually -
None of them got it right (or even identified the right part) even after I cut it down to 10 pages.
Eventually -
Evals are hard for a reason. New post on actually doing them end to end, breaking down the problem, and explaining how we do them at SB
May 12, 2025 at 5:05 PM
Evals are hard for a reason. New post on actually doing them end to end, breaking down the problem, and explaining how we do them at SB
Reposted by Hrishi
> Treat an eval like an aromatic or a condiment - not the meat and potatoes.
A lot of food analogies in the post. I do not recommend reading this on an empty stomach
A lot of food analogies in the post. I do not recommend reading this on an empty stomach
May 9, 2025 at 11:32 AM
> Treat an eval like an aromatic or a condiment - not the meat and potatoes.
A lot of food analogies in the post. I do not recommend reading this on an empty stomach
A lot of food analogies in the post. I do not recommend reading this on an empty stomach
This is the guide I wish I had - didn't hold back.
Everything I know.
Enjoy.
Everything I know.
Enjoy.
May 9, 2025 at 4:12 AM
This is the guide I wish I had - didn't hold back.
Everything I know.
Enjoy.
Everything I know.
Enjoy.
Made this for myself and it's made a lot of people in my life very, very happy
Improvement over SoTA: trying to solve LLM consistency over very long outputs, and output adherence to things like timestamps - where a 500ms change is noticeable.
github.com/hrishioa/ipgu
Improvement over SoTA: trying to solve LLM consistency over very long outputs, and output adherence to things like timestamps - where a 500ms change is noticeable.
github.com/hrishioa/ipgu
GitHub - hrishioa/ipgu
Contribute to hrishioa/ipgu development by creating an account on GitHub.
github.com
April 25, 2025 at 5:12 PM
Made this for myself and it's made a lot of people in my life very, very happy
Improvement over SoTA: trying to solve LLM consistency over very long outputs, and output adherence to things like timestamps - where a 500ms change is noticeable.
github.com/hrishioa/ipgu
Improvement over SoTA: trying to solve LLM consistency over very long outputs, and output adherence to things like timestamps - where a 500ms change is noticeable.
github.com/hrishioa/ipgu
What separates Deepseek is how hardware aware they are in algo design. Perhaps nascency, resource limitations or how they're set up, almost all the recent papers have some reference or awareness around theoretical python research actually meeting at-scale deployments in silicon.
February 22, 2025 at 2:00 AM
What separates Deepseek is how hardware aware they are in algo design. Perhaps nascency, resource limitations or how they're set up, almost all the recent papers have some reference or awareness around theoretical python research actually meeting at-scale deployments in silicon.
Yet another definition for an Agent:
AI with side effects
AI with side effects
February 21, 2025 at 5:22 PM
Yet another definition for an Agent:
AI with side effects
AI with side effects
Vibe coding is crazy
Took an hour or two and made something that can push notes and outputs from Lumentis straight to Notion
Been writing more with Cursor, and pushing it to Notion
Took an hour or two and made something that can push notes and outputs from Lumentis straight to Notion
Been writing more with Cursor, and pushing it to Notion
February 20, 2025 at 6:12 PM
Vibe coding is crazy
Took an hour or two and made something that can push notes and outputs from Lumentis straight to Notion
Been writing more with Cursor, and pushing it to Notion
Took an hour or two and made something that can push notes and outputs from Lumentis straight to Notion
Been writing more with Cursor, and pushing it to Notion
Everytime someone asks me for a good example of company-level writing I point to @flydotio
February 20, 2025 at 6:21 AM
Everytime someone asks me for a good example of company-level writing I point to @flydotio
At the AI Eng Summit starting tomorrow
February 19, 2025 at 3:00 PM
At the AI Eng Summit starting tomorrow
This for me connects to nonverbal thinking in meditative spaces, where you solve hard problems and 'feel' yourself thinking while the internal monologue is silent - no pictures, no sounds, no words.
Is connecting with the self really just feeling the internal latent space?
Is connecting with the self really just feeling the internal latent space?
February 18, 2025 at 5:11 PM
This for me connects to nonverbal thinking in meditative spaces, where you solve hard problems and 'feel' yourself thinking while the internal monologue is silent - no pictures, no sounds, no words.
Is connecting with the self really just feeling the internal latent space?
Is connecting with the self really just feeling the internal latent space?