Henry Wang

383 posts

Henry Wang banner
Henry Wang

Henry Wang

@Henry_Flowgpt

Co-founder FlowGPT & Emochi & Kaon Lab| ex-AMZN building an AI-native content platform with AI-native company

San Francisco Katılım Ekim 2021
5.4K Takip Edilen4.2K Takipçiler
Sabitlenmiş Tweet
Henry Wang
Henry Wang@Henry_Flowgpt·
We (Emochi, with 10M+ users) stopped using benchmark scores for production decisions about 6 months ago. The pattern was consistent: models ranking top on benchmarks underperformed in real usage. Models ranked lower drove better retention. A few notes on what we learned:
English
5
0
8
869
Henry Wang retweetledi
Nebius
Nebius@nebiusai·
10 million users. 100+ models tested every week. More than 1 trillion tokens processed every day. Henry Wang explains why he thinks the next generation of consumer AI won't be won by the best model, but rather by the teams that can iterate and scale the fastest. Listen now: nebius.com/podcast @Henry_Flowgpt @Josh_Liss
English
4
20
178
17.6K
Henry Wang
Henry Wang@Henry_Flowgpt·
I built a @claudeai skill to rate AI proficiency. I think Paxel (paxel.ycombinator.com) is a good start, but it doesn't work for me because of how sensitive my data is — it uploads your code to servers So I made an open sourced one. Repo in the first comment
Henry Wang tweet media
English
2
0
3
157
Henry Wang
Henry Wang@Henry_Flowgpt·
Two things worth putting time developing agent in production 1. Providing accurate context 2. Writing good test cases
English
0
0
0
90
Henry Wang
Henry Wang@Henry_Flowgpt·
Open-sourced Claude Code Web. For people who can’t run Claude Code Client locally, or prefer a browser workspace over the TUI. Runs with local Claude Code CLI or on a remote server via SSH tunnel. Most importantly: you can change it into any skin you like.
Henry Wang tweet mediaHenry Wang tweet mediaHenry Wang tweet mediaHenry Wang tweet media
English
2
1
1
9.6K
Henry Wang
Henry Wang@Henry_Flowgpt·
Small tool that I built to help me give better interview: input: JD + resume - output: questions for each round input: transcript + JD - output: hiring suggestion with detailed reasons (so far 80%+ aligned with my decision)
English
1
0
0
80
Olivia Moore
Olivia Moore@omooretweets·
This is the worst feature Apple has ever made
Olivia Moore tweet media
English
44
2
179
18.9K
Henry Wang
Henry Wang@Henry_Flowgpt·
We (Emochi, with 10M+ users) stopped using benchmark scores for production decisions about 6 months ago. The pattern was consistent: models ranking top on benchmarks underperformed in real usage. Models ranked lower drove better retention. A few notes on what we learned:
English
5
0
8
869
Henry Wang
Henry Wang@Henry_Flowgpt·
One takeaway for 2026: Eval infrastructure will determine product ceiling more than model selection. The question isn't "which model is best." It's "which system learns fastest from real users." Wrote up the full framework: kaon.io/blog/offline-e…
English
0
0
3
224
Henry Wang
Henry Wang@Henry_Flowgpt·
We rebuilt the pipeline on three components: 1. Elo/TrueSkill on real conversations — filters out bad models in hours 2. Conversation-level A/B — not user-level (key distinction below) 3. Reward models trained on behavioral signals — closes the loop into training
English
0
0
2
179
Henry Wang
Henry Wang@Henry_Flowgpt·
Core issue: benchmarks optimize for "correct responses" on discrete tasks. Consumer AI optimizes for "experiences that feel right" across continuous interactions. These objectives aren't aligned. At scale, they actively conflict.
English
0
0
1
162
Henry Wang
Henry Wang@Henry_Flowgpt·
@venturetwins My lesson learned from building Emochi is that the key is not only the model itself but the closed loop eval-feedback-iteration infrastructure. Happy to share more thoughts
English
0
0
1
61
Henry Wang retweetledi
Michael Guan
Michael Guan@dralaska_·
Interviews are broken. resumes mislead. We helped 100k+ people land jobs & scaled to $10M ARR. now we’re rebuilding hiring from scratch. meet WorkTrial AI — where companies see the real work before they hire.
English
198
171
520
654.6K