John Hwang

2.7K posts

John Hwang banner
John Hwang

John Hwang

@ainativefirm

enterprise ai trends. ex-head, vix and variance derivatives, Morgan Stanley

Katılım Aralık 2022
942 Takip Edilen918 Takipçiler
Paper
Paper@paper·
@marcelkargul Try Paper! Our html/css based canvas means agents already speak the same language.
English
0
0
6
202
Kevin
Kevin@itskevin·
My main takeaway from the Delve saga is there’s a shit ton of SOC 2 startups
English
6
1
86
5.6K
Kakashii
Kakashii@kakashiii111·
Today, my work over the years received another validation. The DoJ, together with the FBI, announced charges against three executives of SuperMicro (SMCI) for Conspiracy to Violate the Export Control Reform Act. Since at least early 2024, SMCI's executives built a sophisticated, systematic scheme to illegally divert billions of dollars worth of high-performance AI servers containing restricted Nvidia GPUs to China, using a pass-through company in Southeast Asia to evade U.S. export controls. I was the first to call out this smuggling phenomenon in late 2023, when it was just beginning to pick up volume. I tracked the flow of chips across Southeast Asia, including their final destinations in China, and identified the key participants, SMCI among them, documenting how the sophisticated, systemic scheme operated. What the DoJ described in its charging documents is precisely what I wrote about numerous times: the mechanics of how it works, the routes the GPUs travel from origin to their final destination in China, and the volume of the smuggling, which I estimated at tens of billions of dollars worth of GPUs. SMCI is a significant catch, but it is far from the only player in this smuggling ecosystem. Others remain uncharged. What has been equally troubling is that many companies, including public companies and their executives, have had knowing or willful blindness to this phenomenon, looking the other way in order to hit sales targets and meet Wall Street estimates. Meanwhile, the smuggling network has evolved in parallel with one of the fastest datacenter buildouts in Southeast Asia, involving numerous subsidiaries, local companies, and datacenter operators who have absorbed every lesson from the smuggling playbook and are now working around the clock to build a datacenter empire across the SEA region.
Kakashii tweet media
Kakashii@kakashiii111

SMCI confesses to being Nvidia's partner-in-crime in Singapore sales. Amazing.

English
25
65
377
67.2K
John Hwang
John Hwang@ainativefirm·
@varunram “Agent Native Product”… just a sandbox
English
0
0
1
198
Varunram Ganesh
Varunram Ganesh@varunram·
Cursor's new "Glass UI" is.. just a terminal
Varunram Ganesh tweet mediaVarunram Ganesh tweet media
English
6
0
18
3K
martin_casado
martin_casado@martin_casado·
This is really, really cool. Sort of the file system equivalent of the agent sandbox environments. Imma use it for my setup ...
Mike Freedman@michaelfreedman

Introducing TigerFS - a filesystem backed by PostgreSQL, and a filesystem interface to PostgreSQL. Idea is simple: Agents don't need fancy APIs or SDKs, they love the file system. ls, cat, find, grep. Pipelined UNIX tools. So let’s make files transactional and concurrent by backing them with a real database. There are two ways to use it: File-first: Write markdown, organize into directories. Writes are atomic, everything is auto-versioned. Any tool that works with files -- Claude Code, Cursor, grep, emacs -- just works. Multi-agent task coordination is just mv'ing files between todo/doing/done directories. Data-first: Mount any Postgres database and explore it with Unix tools. For large databases, chain filters into paths that push down to SQL: .by/customer_id/123/.order/created_at/.last/10/.export/json. Bulk import/export, no SQL needed, and ships with Claude Code skills. Every file is a real PostgreSQL row. Multiple agents and humans read and write concurrently with full ACID guarantees. The filesystem /is/ the API. Mounts via FUSE on Linux and NFS on macOS, no extra dependencies. Point it at an existing Postgres database, or spin up a free one on Tiger Cloud or Ghost. I built this mostly for agent workflows, but curious what else people would use it for. It's early but the core is solid. Feedback welcome. tigerfs.io

English
15
4
169
35.7K
John Hwang
John Hwang@ainativefirm·
@varunram Is that true? I may be in a bubble but know a ton of devs still getting paid the same to vibe code for 1 hr a day and getting 400k. Mainly because their companies are run by AI illiterates. Can’t be isolated cases though
English
1
0
1
48
Varunram Ganesh
Varunram Ganesh@varunram·
We're getting close to a future where people make more money from cooking and cleaning dishes at home than they do from their day job at work
Varunram Ganesh tweet media
English
1
0
8
367
John Hwang
John Hwang@ainativefirm·
@varunram Can JPM and Visa just fucking release a cli? Shouldn’t be that hard. Are they this regarded? Stripe shouldn’t be winning this market
English
0
0
0
30
Varunram Ganesh
Varunram Ganesh@varunram·
Agentic Commerce unfortunately is shaping up to become exactly like MCP. Way too many protocols and capital, not many implementing or using it in meaningful ways
English
1
1
14
875
John Hwang
John Hwang@ainativefirm·
@stephenhaney The free tier of Paper was laughably tiny, had to switch to an alternative that launched last week :)
English
0
0
0
89
John Hwang
John Hwang@ainativefirm·
@varunram Your hand crafted UX is my opportunity (to rip off)
English
1
0
1
13
Varunram Ganesh
Varunram Ganesh@varunram·
Google is great at indexing the web and the best at reading and generating images, so they could come up with a really good coding model for frontend designers where most engineers screenshot and paste things into claude code / codex
English
3
0
18
567
Emir Karabeg
Emir Karabeg@emkara·
Introducing Mothership, the first workspace for AI agents. Mothership is the central intelligence layer for your AI workforce. Autonomous agents, fully observable and editable. Check out what Mothership can do below.
English
85
63
416
136K
John Hwang
John Hwang@ainativefirm·
@tokengobbler Hmm, vibekanban is almost there as a factory, just need an agent for managing kanban
English
0
0
1
23
Louis Knight-Webb
Louis Knight-Webb@tokengobbler·
Before we get code factories, we'll get exoskeletons. Agents are still force multipliers. When a human stops working, so do their agents. Factories run 24/7.
Louis Knight-Webb tweet media
English
1
0
2
149
Kevin Kern
Kevin Kern@kevinkern·
codex often marks tasks as finished even when some of the work is still left, so I added a simple task system to my project. It doesn't need much, so a few skills help here. simple: plan -> fix -> review -> fix -> summary code in the comments below
Kevin Kern tweet media
English
3
7
75
8.7K
SkalskiP
SkalskiP@skalskip92·
5 years ago I was using tesseract for OCR - nothing but pain - hard to set up - required model tuning for every font and alphabet - results were so much worst especially for noisy images it blows my mind what we can do this days
SkalskiP@skalskip92

spent most of my day playing with GLM-OCR it's a 0.9B param vision-language model. supports 8K resolution, 8+ languages, and has built-in text, LaTeX, and table recognition modes. awesome! I tested it across different OCR tasks. starting with shipping container serial numbers.

English
11
32
571
44.4K
John Hwang
John Hwang@ainativefirm·
@juecd__ @1Password Does this not trigger security warnings? Also how does this differ from being logged in truly as “you” in chrome let’s say
English
0
0
0
22
catherine jue
catherine jue@juecd__·
We’ve partnered with @1Password to take the next step toward solving authentication for agents. Last month, we introduced Managed Auth: a standardized way for agents to log in and stay logged in across the internet. With this partnership, your agents can now use credentials directly from your 1Password vaults with Managed Auth.
catherine jue tweet media
English
2
2
18
1.1K