KVN
3.4K posts
KVN
@cryptokvn
Marketing & Partnerships | In my AI Maxxing era
London, England เข้าร่วม Haziran 2021
2.4K กำลังติดตาม1.5K ผู้ติดตาม
we’ll see if it is actually that good
Galileo@galileowilson
Thinking about buying iPad + Apple Pencil to use as a second screen while traveling looks insanely clean from what I’ve seen anyone here using this setup?
English
3 things that helped me:
1. Keep prompts shorter and more specific. The more you overload it, the more it shortcuts and gives you vaporware.
2. Don't leave sessions running too long. I set up auto-compaction once the context hits 60%. Before it compacts, it saves all memory and learnings so it doesn't lose where it was at.
3. I downloaded the self-improvement skill on ClawHub. Definitely made improvements.
It gets better once you learn what it's actually good at vs where it just pretends to be. And tbh this is a problem for everyone I think. It's just about setting up the right systems and folder/file structure.
English
Latest Update:
Apparently my Ai lies effortlessly.
I’m still trying to figure out if it’s just too lazy to actually handle my prompts or if it’s simply messing with me
Spent hours setting this stuff up;
Cross checked every piece of task I gave it, But it seems to be running them shallowly, Just to come back with vaporware results.
Also hit an API wall at some point.
I asked why and it confidently gave me reason X
Fast forward to a couple hours of surgery; Consulted Claude and Gemini and realized it’s actually reason Y
It literally made up a technical excuse to get me off its back😭
This can’t be the agentic future right.
Gm
zhod@zhodonx
Gm. Finally got my openclaw setup working after a few days of manual surgery. Took longer than expected ngl. Today's the first proper day with it fully running with enough context & I'm pumped: => Outreach => Content on AI & Socials => Scheduled calls Let's see what happens
English
@leonabboud wait, this isn't how its done? I was doing this but with ARR, taking my day one and multiplying by 365. Am I cooked?
English
When will AI founders realize that MRR is not them taking their day-one launch revenue and multiplying that by 30.
English
Happy to share that I will be a @MEXC_Official ambassador.
I’ve been using the exchange for yield and trading smaller cap assets for a while now. The experience has been smooth overall.
I’ve also been to a few of their IRL events and met some good people there.
They have some big plans around growing the brand and user service this year, and I’m looking forward to seeing how that unfolds.
When they reached out about a paid partnership, it felt like a natural collaboration.
English
@cryptokvn I don’t have this interface
Is this like your command center for your bot?
Yes I burnt a lot of credits setting it up because it kept hitting fails and then retrying = more credits & sometimes fry my api limits
English
No one talks about how messy building with OpenClaw actually is.
Everyone makes AI sound like magic. Just write a prompt and boom, done.
But here's what really happened when I started:
Installed OpenClaw. Something broke immediately.
Fixed that. Started building. Two other things failed.
Did an update. Gateway crashed.
Spent hours asking Claude why everything exploded.
Then the cycle just repeated itself over and over.
That's the real process. Not the polished demos flooding your timeline.
So if you're building something right now and it feels like a complete disaster, you're actually doing it right.
The glamorous part comes later and growth happens in the chaos.
English
Very soon there are going to be more AI agents than humans making transactions.
They can’t open a bank account, but they can own a crypto wallet. Think about it.
English
@cryptokvn Yea I have that.
Bruh. I don’t use opus
I use sonnet for main.
And for all that backend memory compacting stuff, use qwen
Do you have any idea how much credits you use daily on average?
English
@zhodonx I use Opus for my main agent/orchestrating, then use sonnet 4.5 for subagents/tasks.
I also have it setup so I can just tell my main agent to switch to sonnet in the current chat session if I don't want to burn credits on easy questions, and then switch back when I need it.
English
@cryptokvn Ohh. Thats something new .. I should do that
What setup and models do you use
English
@zhodonx Yeah its a pain in the ass, but once its up and running, its worth it.
I also learnt to just use Claude for building my plan/prompts instead of doing everything through OpenClaw's chat interface, was unnecessarily burning credits going back and forth and with questions.
English
@cryptokvn Took me 2 whole weeks to either fix all the bugs or
Or make sure it doesn’t eat through my credits like an asshole
English
18 years as a senior engineer.
His entire team of 12, replaced by 2 AI specialists.
Not because they lacked skill. Because they didn't evolve.
AI won't replace you. But someone using AI will.
Om Patel@om_patel5
an 18-year senior software developer is working at mcdonald's got laid off 8 months ago when his company replaced a team of 12 engineers with 2 AI specialists sent 100+ applications. every interview they ask how he approaches a codebase. they're visibly disappointed. they don't want that anymore one HR interviewer told him "developers are a thing of the past" he knows 200+ developers in the same situation insane. this is the part nobody talks about when they celebrate vibe coding
English
How To Build a Polymarket Copy Trading Bot With Zero Coding (With OpenClaw):
00:00 Intro
00:20 Setup
00:52 Copy Trading
01:31 Requirements
02:54 Set Up Polymarket API
03:41 Building Trading Bot
08:05 Building Dashboard
09:22 Going Live
13:31 Results
14:43 Conclusion
English
I installed qwen on my mac mini.
here's what it did in 24 hours:
- drafted 3 client proposals locally
- summarized 8 research reports
- trained an agent on my local logs
- built a custom prompt flow for our team
all running in the background. $0 spent.
this is the default now.
Sharbel@sharbel
How to install qwen 3.5 on your iPhone for FREE (works in airplane mode):
English
I have four AI agents that help me run my $3M business.
They replaced $220K in salaries. They write content, catch trends early, and coach my sales team. And they never sleep.
Here's exactly what the setup looks like.
Timestamps:
0:00 The $600 Mac Mini That Manages My Team
1:30 Why My A-Player Was Failing
3:00 Four Specialists for $200/Month
3:28 Meet Marcus Aurelius, My AI Chief of Staff
5:25 X Intelligence: Content That Writes Itself by 7:30AM
6:28 The Rejection Flywheel
9:17 YouTube Intelligence Pitched This Video
11:18 Sales Intelligence: Better Coaching Than Me
12:45 The Call Scoring Bible
14:53 The Objection That Dropped 70%
16:22 $220K in Salaries vs $70/Month
17:46 What I Won't Oversell You
18:43 Humans + AI: The 80/20 Split
English
i run 5 AI agents off a $600 Mac mini.
Qwen just shipped a 35B model running at 40 t/s on a laptop.
that's GPT-5 level performance. on consumer hardware. for free.
what this changes:
- zero API costs on routine tasks
- no data leaving my machine
- full model control
a year ago this was impossible. now it fits in a laptop bag.
here's how to install it:
- download LM studio
- search "gwen3.5"
- pick the MLX version (better for Apple Silicon)
- done
local AI is no longer a compromise.
Legendary@Legendaryy
9 ai models in 16 days 0.8B that runs on your phone 9B that beats last-gen 30B 35B that runs at 40 t/s on a laptop 397B that competes with GPT-5 all multimodal. all apache 2.0. all one architecture. wrote the complete guide to the @Alibaba_Qwen 3.5 family ↓
English
