Boxmining

22.7K posts

Boxmining banner
Boxmining

Boxmining

@boxmining

Building fast on with AI until it breaks | Web3 VC

Web3 参加日 Mart 2017
6.1K フォロー中177K フォロワー
固定されたツイート
Boxmining
Boxmining@boxmining·
Stealth launching my AI channel Real AI and OpenClaw use cases without the fluff. @BoxminingAI" target="_blank" rel="nofollow noopener">youtube.com/@BoxminingAI
English
19
2
91
30.8K
Boxmining
Boxmining@boxmining·
Qwen 3.7 Max (@Alibaba_Qwen) just changed the AI agent game. We gave it complex real-world coding tests that take top models 3 hours to finish. It crushed them in under 9 minutes with a 100 percent success rate. No context rot, pure endurance. Is this the most powerful AI coding backbone ever built?👇
English
10
1
45
2.8K
Boxmining
Boxmining@boxmining·
🚨 OpenHuman has MAJOR potential, but also MAJOR bugs. It has some real strengths, direct API connections to major apps without relying on computer use. But there are bugs worth knowing about before you dive in @tinyhumansai Watch the full breakdown 👇
English
7
0
16
1.8K
Boxmining
Boxmining@boxmining·
This is rare.. 5K users. 4K daily active users. $0 in marketing growth costs. $0 in VC funding In an ecosystem where everyone's optimizing for rounds… Someone's just optimizing for product. Stay human. Indeed.
Steven Enamakel@senamakel

OpenHuman has crossed 5000 users and over 4000 daily active users. And has spent 0$ in marketing and 0$ in VC funding. We are fixing bugs, doing massive improvements, massive features.. Stay Human! We're making it a whole whole lot better. Thanks for being early ❤️

English
3
0
10
2K
Boxmining
Boxmining@boxmining·
Hot take: Hermes Agent @NousResearch new Skill Bundles are less about saving clicks and more about making agents behave consistently. Instead of hoping your agent picks the right tools, you pre-package the workflow: research, code review, backend dev, whatever. This is how AI agents become less random and more useful.
English
6
0
69
3K
Boxmining
Boxmining@boxmining·
The underrated part of @NousResearch Hermes agent skill bundles is not speed, it is reliability. Agents are probabilistic, so if your workflow needs 5 skills, there is no guarantee the agent picks all 5 unless you make that structure explicit.
English
2
1
15
822
Boxmining
Boxmining@boxmining·
Hermes agent skill bundles are a small feature with a big workflow upgrade. Instead of calling skills one by one or hoping the agent chooses correctly, you can bundle the right skills and load them with one slash command. Watch the full setup guide on @NousResearch and start building cleaner AI workflows.
English
5
4
41
2.8K
Boxmining がリツイート
Humanity
Humanity@Humanityprot·
Humanity chain is now supported on Bybit and Kucoin! Send your tokens directly, no bridging needed. More coming. Watch this space 👀
English
67
75
300
37.2K
Boxmining
Boxmining@boxmining·
Hermes Agent (@NousResearch) just got a serious speed upgrade. Faster startup, cleaner chat loop, lazy compression, and terminal/file tools that finally feel instant. Small engine changes, huge UX difference. If you use Hermes for coding workflows, you need to see what changed.
English
6
3
52
4.1K
Boxmining
Boxmining@boxmining·
The new AI tools you are using right now don't stay open forever. At first, nobody gets it. Then the early movers build. Then the corps take over. That's the pattern. Crypto. AI. Every shift before it. Your move? Use it or watch it pass. Build now. Thank yourself later.
Autism Capital 🧩@AutismCapital

Hot take: There is a limited window of time, like any new industry, where a new tool appears and people don't understand it fully yet, and it's powerful enough to change your entire life, but not so powerful it's been completely taken over by corporate interests yet. And when this tool appears you have two choices: use it to skyrocket your life and take advantage of this TEMPORARY moment, or resist/hate/be angry/complain/, do nothing, and the moment is gone forever. These brief moments in time are when shifts in the game board happen. It happened with Crypto, its happening with AI, it happens with any NEW thing. These shifts in the game board allow people who came/come from nothing, to elevate, using nothing but DEMOCRATIZED tools, moving first, their creativity, and execution. Our take is that instead of poo poo-ing the tool, USE the tool, improve your life, try applying the tool to the industries/things you know best, get creative, and catapault your way into a new class. You'll fine the sense of fulfillment you get as you're building, growing, expanding, learning, elevating your life, to FAR OUTWEIGH any feelings of superiority you get by hating and rejecting the tool. If you don't use this time, VERY likely chance in the future, whether 5 years, 10 years, when you are older, have more family, more weight/responsibility, and the realities of life are much more tangible, you'll wish you had acted more during this time. You don't want to be filled with regret because you chose to hate/dismiss, instead of build/grow, which ONLY helps you and the ones you love. TLDR? Instead of hating/dismissing a new tool, use this BRIEF WINDOW of time to elevate/improve your life. The window is temporary and if you don't use it because you want to please a small niche of people who don't really care about your success anyways, you may very highly regret that decision. Disclaimer: Post written with care and consideration and in a good spirit. Please don't be hateful for no reason. This is to benefit those who these words get through to.

English
7
0
9
1.1K
Boxmining
Boxmining@boxmining·
The interesting part of this Hermes Agent update is not one huge breakthrough. @NousResearch It is a bunch of small speed wins stacked together - lazy loading - fewer redundant calls - smarter polling -less startup overhead That is usually how good software gets fast.
English
3
1
16
740
Boxmining
Boxmining@boxmining·
MiniMax Mavis (@MiniMax_AI) feels like one of the first AI agent tools that actually understands beginners. Instead of making you wire everything together, it gives you agents, teams, scheduled tasks, and a verifier in one place. Is this the direction AI tools should be going?
English
2
0
18
1.3K
Boxmining
Boxmining@boxmining·
The most interesting part of @MiniMaxAgent Mavis is the “devil’s advocate” agent. Most AI agents can produce work, but they’re often bad at questioning their own output. Having a separate verifier feels like a small feature that could make a huge difference. Big win for @MiniMax_AI
English
6
3
29
5.2K