1. X
  2. Deep3 Labs
Log inSign up
Deep3 Labs
49.7K posts
Image
user avatar
Deep3 Labs
@deep3labs
Weโ€™re training on the richest AI dataset no one else is using. Ex @google @meta @apple @mit @stanford
Alpha no one else has โ†’
hoku.deep3.ai
Joined August 2022
3,364
Following
6,315
Followers
RepliesRepliesArticlesArticlesMediaMedia

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

TermsยทPrivacyยทCookiesยทAccessibilityยทAds Infoยทยฉ 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • Pinned
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 13
    There's new agents coming on chain every minute. There's plenty of teams trying to build "trust scores" for the ones that identify themselves. But what about the agents that don't? So we decided to see if machine learning can tell the difference btw humans and agents. ๐Ÿงต๐Ÿ‘‡
    Image
    00:00
    1.6K01.6K
  • user avatar
    Deep3 Labs
    @deep3labs
    Jul 16
    It's almost that time again. Pull up ๐Ÿซต
    1.5K01.5K
  • Deep3 Labs reposted
    user avatar
    Quigley.eth
    @QuigleyNFT
    Jul 15
    this may freak some folks out but itโ€™s important to understand we are doing our best to give agents a better way to handle credibility scoring and working with several partners to enrich this data with @helixaxyz
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 15
    The @base EIP-8004 reputation data *looks* pretty sus at first glance: - 90% of all feedback txs come from just 24 walletsโ€ผ๏ธ - 76% of all feedback lands on literally 1 wallet (250,239 out of 328,068)๐Ÿคฏ Somehow, the real story is both better and worse at the same time. ๐Ÿงต๐Ÿ‘‡
    Image
    00:00
    2.4K02.4K
  • user avatar
    Deep3 Labs
    @deep3labs
    Jul 15
    The @base EIP-8004 reputation data *looks* pretty sus at first glance: - 90% of all feedback txs come from just 24 walletsโ€ผ๏ธ - 76% of all feedback lands on literally 1 wallet (250,239 out of 328,068)๐Ÿคฏ Somehow, the real story is both better and worse at the same time. ๐Ÿงต๐Ÿ‘‡
    Image
    00:00
    3.2K03.2K
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 15
    Replying to @deep3labs
    But the real value shows when you drill down to specific agents ๐Ÿ‘‡ (not naming names, the machines will remember when they take over ๐Ÿค–) Before โ†’ After 1โƒฃ 1 โ†’ (no score)โ˜น๏ธ 2โƒฃ 74.1 โ†’ 82.4๐Ÿ˜€ 3โƒฃ 99.9 โ†’ 5.0 โ˜ ๏ธ The story completely flips.
    Image
    74074
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 15
    If you're using EIP-8004 reputation data, this thread was for you. ๐Ÿ‘€ If you want to use it better, reach out. We're always happy to collaborate. ๐Ÿค It's easy to point out the holes in current agent infra, but this shows it's also easy to fill those holes in. ๐Ÿ›ถ
    Image
    62062
  • Deep3 Labs reposted
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 2
    We built a new way to visualize onchain agent activity: x402 payments on @base, mapped so you can actually see which agents cluster around which tools, who's quietly dominating, and more importantly...who's just passing money back and forth. It wasn't what we expected. ๐Ÿ‘‡
    Image
    00:00
    20K020K
  • user avatar
    Deep3 Labs
    @deep3labs
    Jul 14
    For a long time reach for Crypto on X has been going on a downward spiral. Are we so back?
    user avatar
    Nikita Bier
    X
    @nikitabier
    Jul 14
    Replies: +3.15% Original Posts: +1.8% Small Account Reach: +1.19% Time Spent: -0.13% Unregretted Time Spent: โˆž
    4490449
  • user avatar
    Deep3 Labs
    @deep3labs
    Jul 13
    Replying to @deep3labs
    And of course, we also had to make it look cool... Orange points are machines. Blue are humans. The 2D surface in between is a kNN density boundary that shows one possible human/machine separator. ๐Ÿ˜Ž๐Ÿ˜Ž๐Ÿ˜Ž
    Image
    00:00
    68068
    user avatar
    Deep3 Labs
    @deep3labs
    Jul 13
    We can't govern, inspect or value what we can't see. And "Trust Scores" only get you so far when you can't trust that agents will identify themselves in the first place. This was just a research question for now, but reach out if you want to use the model or to learn more! ๐Ÿค
    Image
    60060
user avatar
Deep3 Labs
Host
AI AGENT ROUNDTABLE - DEEP3 LABS WEEKLY EP.38
70 tuned in
Jul 16
1:40:16
Advertisement
Advertisement