Jeff Lin
@jefflinshu
Followers
75
Following
30
Media
17
Statuses
60
Enable auto mode agentic product manager. Stay curious and keep creating.
Hangzhou
Joined August 2017
The most dangerous and useful thing about LLMs is the same: They can argue brilliantly for almost anything. What feels like “validation” is often just bias reinforcement with better wording. Best use: Have it support your view. Then have it dismantle it. If both sides sound
0
1
1
What's your first impression of Grok? Most people say: the images. The voice. The X integration. Mine? Speed. We deployed Grok as an AI customer service agent. Users ask a question — they need an answer before anxiety sets in. We tested a lot of models. Most of them made
1
0
1
If you had to name the defining product of 2026, what would it be? For me, it's not even close. It's the lobster. 🦞 @openclaw Here's why I think it's the most important software of our generation: Before OpenClaw: AI lived in apps. You had to go to it. After OpenClaw: AI
1
3
7
Lovable is building the world where every person is an AI engineer — and their team lives it. 54 days. 1B tokens/minute. ARR $300M → $400M. March 8 free for everyone — 500k lines of code written while you read this tweet. General Tasks, penetration testing, Plan Mode, automated
0
0
2
OpenAI is moving fast. Codex is moving faster. And GPT-5.4 made it faster still. 54 days. A new model architecture, Automations GA, Subagents, native Xcode integration — and a client that somehow got more elegant with every release. Feb 2: macOS app, 500K downloads day 1.
2
0
2
Kiro didn't just ship an AI IDE. They shipped a new way to think about building software. I spent weeks reading how Kiro writes its three documents — requirements, design, tasks. Not because I had to. Because it changed how I structure my own thinking. Two underrated details:
1
0
3
I was one of the earliest beta users of @augmentcode. Then I paid for it. Here's why. Most AI coding tools optimize for one thing: generating code faster. Augment went a different direction — they built a Context Engine that actually understands your entire codebase.
1
0
4
Hermes Agent doesn't just run tasks — it learns from them. Creates skills, remembers you, recalls across sessions. An agent that grows. 26 days. Four major versions. 10k GitHub stars. A 79k-word novel written by the agent itself. My colleague @nemoaigc (Nemo) built an AI tavern
3
8
110
Cursor redefined the Tab key. That alone changed how I code. But the real unlock is Multi-Agent — you stop managing code, you start managing outcomes. @cursor_ai has been quietly shipping thisvision for months. Agent Harness, Remote Control, Composer 2, Figma integration,
0
0
2
Stitch launched last week and honestly our team hasn't stopped talking about it. The design DNA thing is real — we plugged it into an actual project and it just works. Pair that with @GoogleAIStudio's new full-stack mode and you can go from zero to deployed app without leaving
0
1
6
Been hearing people rave about @AmpCode for months. Finally tried it myself. The design is genuinely elegant — feels like someone thought hard about what a coding agent UI should be, not just what it could be. Small team. Deeply focused. @sqs @beyang @thorstenball and crew have
2
2
36
Nemo you’re literally a genius.
0
0
1
thx Greg, I try to launch my new app soon...
do yourself a favor learn to get good with ai pick a niche build distribution build apps or agents as a service make something so useful then expand around that behavior 99% of people will read this and do nothing 1% will read this and do their future selves a favor i don't
0
0
1
Love this idea. I’ll try it out and then polish Elon Musk’s profile.
The biggest time-waster at most tech companies isn't meetings or Slack. It's trying to predict what your leader wants. In my latest post, I break down how a VP at Meta solved this by building an AI skill that reviews docs in his voice: → What the /exec-review skill does →
3
0
2
Kinda feels like I should move to SF.
STATE OF THE PRODUCT JOB MARKET IN EARLY 2026 In spite of the headlines about layoffs and AI taking jobs, we’re actually seeing a lot of promising signs in tech hiring, and some interesting new trends: 1. PM openings are at the highest levels we’ve seen in over three years 2. AI
0
0
0
can't waiting for it.
We're starting "What We Shipped," a monthly stream where we'll share our latest tips and talk about new Claude Code releases. The first one is April 7th, come hang out with @dmwlff and me!
0
0
1
This isn't just an Elon experiment anymore. The same system works for any public figure on X — investors, founders, politicians, athletes, anyone with a following list and a pattern of attention. Map their network. Score their signals. Know what they care about before they tweet
It opens with: Good morning, Elon. Here's your intelligence briefing for today. I'm not sending it to him. Not yet. But the exercise taught me something more valuable than the briefing itself: when you can map someone's information environment this precisely, you stop
0
0
2
It opens with: Good morning, Elon. Here's your intelligence briefing for today. I'm not sending it to him. Not yet. But the exercise taught me something more valuable than the briefing itself: when you can map someone's information environment this precisely, you stop
I mapped his following list, tagged every account by domain — AI, Space, Tesla, politics, crypto — and built a weighted scoring model on top of his recent retweet behavior. The logic: the more recently he amplified someone, and the higher the engagement, the more that person's
0
0
3
I mapped his following list, tagged every account by domain — AI, Space, Tesla, politics, crypto — and built a weighted scoring model on top of his recent retweet behavior. The logic: the more recently he amplified someone, and the higher the engagement, the more that person's
Here's the thing about reaching Elon: he owns the platform, so money doesn't work. He gets hundreds of thousands of mentions a day, so shouting doesn't work either. The only real path in is understanding what he's actually paying attention to — and meeting him there. Elon
0
0
1