Daniel Yurkin

0

1

26

Chan@chanpark_xyz·1h

@danyurkin @speedrun @atomic_chat_hq Love the local-first angle here. What’s the main use case you’re optimizing for first: personal memory/workflow, agent runtime, or cross-device collaboration?

English

0

1

14

Daniel Yurkin@danyurkin·1h

just applied to @speedrun with @atomic_chat_hq! we're building local ai you can actually use - for humans and the agents they run. open-source, macos, windows, ios fingers crossed 🤞

a16z speedrun 🧊@speedrun

English

0

3

71

Daniel Yurkin@danyurkin·3h

hi andrew! previously led growth at atomic wallet (15M downloads) - built partnerships with solana, bnb, base and tron ecosystems. now building @atomic_chat_hq - local AI runtime and agent for privacy-conscious users. 20k+ downloads in 1.5 months, default local backend for 5 agent frameworks - opencode, goose, hermes workspace, nanoclaw, openclaude

English

5

129

Daniel Yurkin@danyurkin·4h

@adrielyong @speedrun hey @adrielyong! building @atomic_chat_hq - local AI runtime and agent for privacy-conscious users. already applied but would love your thoughts! sent you a DM

English

42

adriel@adrielyong·1d

been referring many folks to the @speedrun team over the last few weeks! Sorry if I missed some DMs, feel free to reply below to ping me on what you’re building someone also just made a directory with everyone associated with speedrun: noticed.so/speedrun great resource

English

64

4

124

10.9K

Daniel Yurkin@danyurkin·4h

@andrewchen made it 🏁 applied with @atomic_chat_hq

English

2

50

Daniel Yurkin@danyurkin·4h

@emilybenn12 hey Emily! applied with @atomic_chat_hq ✅ excited to see what’s next

English

27

Emily Bennett@emilybenn12·11h

fyi - last call to apply for a16z speedrun - deadline is tonight at MIDNIGHT!

English

14

0

56

2K

Daniel Yurkin@danyurkin·11h

@Tocelot @speedrun just submitted @atomic_chat_hq! thanks @Tocelot 🤝

English

1

41

Jon Lai@Tocelot·15h

final day to apply for a16z @speedrun! there's never been a better time to build. every app, dev tool, infra layer, and vertical is being reinvented. users and companies are more open to trying new products than ever if you've thought about taking the leap to start a company or raise $, here's what's on the other side: - the a16z network - an operating team that gives you unfair advantages in recruiting, GTM, marketing, and fundraising - partners who've all been builders, in it with you for the long term - a community of 400+ fellow founders, many of whom are repeat founders of billion$ companies, olympiad medalists, PhDs at the top of their field etc if you're bold enough to build a generational company, we want to partner with you apps close tonight at midnight 👇 speedrun.a16z.com/apply?utm_sour…

English

32

1

114

8K

Daniel Yurkin@danyurkin·1d

for sure: Research and compare two GitHub repositories: Hermes Agent and OpenClaw. Step 1: Collect GitHub stats - stars, forks, contributors, creation date, latest release. Step 2: Get full star history for both projects. Step 3: Find key events that caused growth spikes - viral moments, launches, press coverage. Step 4: Build an HTML dashboard with timeline chart, stats table, milestones, and news section. Save as hermes_vs_openclaw.html and open in browser.

English

1

52

SilentMuchacho@Sank_2o11·1d

@atomicbot_ai Can you share the prompt?

English

0

1

11

Daniel Yurkin retweetledi

atomicbot.ai@atomicbot_ai·2d

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

77

45

654

579.1K

Daniel Yurkin@danyurkin·1d

@populartourist hey @populartourist sharp read! you nailed it - openclaw is built for “get the answer now” hermes for “get better at this over time”, different tools for different tasks

English

1

36

wd 🔺@populartourist·2d

Interesting outcomes with different flavours for different strokes. Openclaw favoured direct pipeline execution, studied fewer sources to get an outcome which directly impact on wall clock and token spending. Hermes inclined into more research depth and then engineer it. More sources, more information - hence the higher time and tokens. On top, Hermes added a small overhead tax by writing a SKILL.md - an important feature for repeatable, more efficient outcomes. If the research quality matters, Hermes is a preferred choice, especially if it's repeated work - efficiency compounds with SKILL.md If sufficient information matters, and tasks don't repeat often, Openclaw potentially wins here, as Hermes prefers RL and optimise future repeated-turns. Importantly, this is one small example. Other use-cases might differ.

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

0

8

873

Daniel Yurkin@danyurkin·1d

@ahmedmoro_ 🔥

QME

6

Ahmed Moro@ahmedmoro_·2d

hermes agent evolving

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

0

2

92

Daniel Yurkin@danyurkin·1d

@ivanfioravanti thanks for sharing us @ivanfioravanti 🤝

English

0

1

21

Ivan Fioravanti ᯅ@ivanfioravanti·1d

This is the kind of benchmark we need to improve things! Well done 💪🏻

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

3

2

28

4.2K

Daniel Yurkin@danyurkin·1d

@Teknium @atomicbot_ai ymmv across different tasks though

English

1

698

Teknium 🪽@Teknium·1d

@atomicbot_ai I mean.. did you leave openclaw's default of no reasoning and hermes' of medium reasoning...?

English

15

2

315

29K

Daniel Yurkin@danyurkin·1d

@sunnypause @atomicbot_ai you can download the app from atomicbot.ai or run it in the cloud

English

3

61

Sunny@sunnypause·2d

@atomicbot_ai how do i get that webui chat to work?

English

2

0

2

2K

Daniel Yurkin@danyurkin·2d

@f0undry_labs @the_great_hug0 let's goooo

English

1

15

Foundry@f0undry_labs·2d

@the_great_hug0 @danyurkin 👍

QME

0

3

60

Foundry@f0undry_labs·2d

Daniel Yurkin is now on Foundry! A product-driven leader and growth strategist who scaled Atomic Wallet and is now pioneering the next generation of AI-powered tools within the Atomic ecosystem. More info about @danyurkin 👉 foundrylabs.fyi/founders/danie…

English

12

253

Daniel Yurkin@danyurkin·2d

@alaamurad @svpino 😬 which model was it running under the hood?

English

0

15

Alaa Murad@alaamurad·2d

@svpino so far, Hermes sucks, I rather vibe code the problem myself, it spent 60$ and a day failling to parse simple page. Those tools are for simi technical people, for coder, just vibe code

English

0

1

258

Santiago@svpino·2d

I still don't know which one is better: OpenClaw or Hermes. Every time I ask somebody, I get a different answer. This is looking like a "Claude Code vs Codex" discussion: everyone has a favorite. If anyone has a strong opinion, please let me know.

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

77

8

127

35.1K

Daniel Yurkin@danyurkin·2d

@josephathomas @svpino 100% Joseph, same on my end - hermes is just snappier out of the box, openclaw takes real time to configure and dial in. honestly hermes is my pick

English

21

LabelGuy@josephathomas·2d

@svpino I think you got a look at the underlying model. I think both harnesses are decent, but from my experience, Hermes has been far more stable. It handles updates, so I'm spending more time being productive than fixing and configuring.

English

0

1

228

Daniel Yurkin@danyurkin·2d

@RoyStory_4 @testingcatalog thanks Roy, solid point! what does your actual agent workflow look like - what do you delegate to them? building in this space, would love a quick call to hear how you use it in practice if you’re open

English

81

Roy@RoyStory_4·2d

@testingcatalog Most ppl woudlnt use Hermes’ this way. We would us Hermes’ to delegate codex or open code. This is like needing a screw driver but using a hammer lol…..

English

2

0

7

2.8K

🚨 AI News | TestingCatalog@testingcatalog·2d

Atomic Bot put Hermes and OpenClaw head-to-head on the exact same task, running the same model (Qwen 3.6 35B) with the same goal: analyzing GitHub history, mapping growth spikes, and shipping a live dashboard in the browser. Key metrics to watch for 👀 > Time to complete the task > Tokens spent > Quality of the final result (dashboard in browser)

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

English

27

31

470

138.4K

Daniel Yurkin@danyurkin·2d

this one: Research and compare two GitHub repositories: Hermes Agent and OpenClaw. Step 1: Collect GitHub stats - stars, forks, contributors, creation date, latest release. Step 2: Get full star history for both projects. Step 3: Find key events that caused growth spikes - viral moments, launches, press coverage. Step 4: Build an HTML dashboard with timeline chart, stats table, milestones, and news section. Save as hermes_vs_openclaw.html and open in browser.

English