MinChi @minchi

Building @CoinFello AI agent for onchain finance • prev @500globalVC • fellow @salzburgglobal @kernel0x minchi.co New York, NY Joined October 2019

Tweets

3K
Followers

1K
Following

982
Likes

4K

MinChi @minchi

7 hours ago

@emilylai Can you please try ordering food delivery in a high speed train? 🥹

1 0 1 58 0

View Details

MinChi @minchi

9 hours ago

I use Siri the most when I’m cooking usually to set a timer when both hands are busy. But now that I’ve been testing other personal agents I wish I could say “Hey Instinct" "Hey Bot" instead

0 0 0 44 0

View Details

MinChi @minchi

10 hours ago

@pratikg what a view Pratik!!

0 0 0 49 0

View Details

MinChi @minchi

14 hours ago

@anshnanda 👀👀

0 0 0 8 0

View Details

MinChi @minchi

15 hours ago

@uttam_singhk Congrats! I know exactly that feeling 👏🏻

0 0 0 27 0

View Details

MinChi @minchi

20 hours ago

@kashvi Ramona!!!!

1 0 1 40 0

View Details

MinChi @minchi

20 hours ago

X timeline over the weekend: jev jev jev

0 0 4 76 0

View Details

MinChi @minchi

20 hours ago

@taojukfc @CoinFello 🔥 🔥 🔥

0 0 0 6 0

View Details

MinChi @minchi

2 days ago

@juliafedorin And I want all the people calling cottage cheese pancakes the “Hailey Bieber pancakes” to try Syriniki!

0 0 0 98 0

View Details

MinChi @minchi

2 days ago

@therollupco @aave @StaniKulechov not defi anymore, onchain finance!

0 0 0 86 0

View Details

MinChi @minchi

2 days ago

@Lingersj Tacos Domingo in Ukranian Village! 🌮

1 0 2 39 0

View Details

MinChi @minchi

2 days ago

I found my favorite flour tortilla based taco spot in nyc

2 0 8 212 0

View Details

MinChi @minchi

4 days ago

@forgebitz would they make similar models in-house instead of collaborating with TypeSafe?

0 0 0 19 0

View Details

MinChi @minchi

4 days ago

adding a paper trading mode has become so much easier with Jev

Tommy Jepsen @tommy_jepsen

4 days ago

Jev trading the danish stock market for all of 2025(239 trading days). Sentiment-based on market data, news articles, wikipedia & google trends. 8.1 million tokens for $0.32 🤯

37 23 850 124K 1K

0 0 0 184 0

View Details

MinChi @minchi

4 days ago

@andruyeung and I was proud to have 15 invites 😅

0 0 0 277 0

View Details

Jev is simply put an AI decision-maker that makes calls based on probability To test it, I wanted to build something quick and useful for me: a prototype where I enter the name of any NYC restaurant/coffee shop and it pulls 5 google reviews and opinions scattered across reddit (eg. from subreddits like /nycfood) It compares them using a set of parameters and then tells me whether the place is legitimately worth trying or just overhyped. I built this because I am officially done falling for overhyped places in this city 🫣 And! it genuinely felt like a natural use for Jev because restaurant reviews are noisy, full of paid content, and contradictory, so the real task is weighing imperfect signals and making a probabilistic call

0 0 0 111 0

View Details

MinChi @minchi

4 days ago

@davidim Oh wow didn’t know 👀

0 0 0 206 0

View Details

MinChi @minchi

4 days ago

@junha_yang Do you think the token subsidies by the govt has an effect on this data?

0 0 0 26 0

View Details

MinChi @minchi

4 days ago

@SophiaChen2003 Did you connect it to your analytics to separate the actual users drop off vs. where grokbot got stuck here?

1 0 0 24 0

View Details

MinChi @minchi

5 days ago

Earlier this year when Claude managed agent launched, I built a "competitive intel analyst" agent and fed it with extensive context of our product, what I want the agent to compare and contrast and when the agent was done the report was shared to the entire team in a Slack channel I named competitive-analyst. The experience was far better than building a custom n8n but it still didn't solve the biggest bottleneck: me. Because I had to trigger the managed agent every time by opening a new session it was far from being an autonomous agent. The past couple of days I've been testing GrokBot in my workflow and tried the "stalk bot" (mentioned in their livestream yesterday) and I'm impressed that the bot goes and finds competitors, signs up for their products autonomously, and then reports back: - what they shipped - what they're posting on X - who they're hiring - what they're actually prioritizing and because the bot has access to its own computer it went through our competitors' onboarding process, compared them against ours, and pointed at where users would drop off. It even told me that: "If you want, next I can walk the live login → connect → first automation screens on video and mark exact drop-off steps." soooo cool! We've been revamping our onboarding behind the scenes, so the timing is absurd. Any other cool grokbot templates others have created for competitor analysis?