
EX $SHIB Developer | Founder & Cryptocurrency Investor Expert in Full-Stack Development, AI, Machine Learning, and Blockchain Systems
Zurich, Swiss 🇨🇭
Joined May 2014
- Tweets7.5K
- Following5.7K
- Followers44.3K
- Likes8K
Edinson Carranza retweeted
The AI startup bubble is getting ridiculous.
Same APIs. Same models. Same features. Different logos.
Everyone is building apps. Few are building sustainable businesses.
By 2030, AI could make many of today's apps unnecessary.
Edinson Carranza retweeted
Unpopular opinion:
Most "AI startups" are just a prompt, a Stripe link and a landing page.
And some of them are making more money than funded startups with 20 engineers.
Is that a problem or the future?
Edinson Carranza retweeted
Two things I keep hearing from builders:
"AI will replace devs."
"Crypto is dead."
Both get said every year.
Both keep shipping.
Which one do you think ages worse?
Edinson Carranza retweeted
The best builders I follow have one thing in common:
They ship before they feel ready.
Not perfect. Not polished.
Just live, then better.
What's the thing you've been "almost ready" to ship for too long?
Edinson Carranza retweeted
Nobody talks about the real cost of building with AI.
It's not the $20 subscription.
It's the credits you burn when the agent:
- retries the same fix 6 times
- rewrites files you didn't ask for
- reads your whole repo to change one line
AI isn't cheap. It's just billed later.
What's the most credits you've burned on a single bug?
Edinson Carranza retweeted
Building a startup and looking for more visibility?
I'm open to collaborating with founders and innovative projects.
Let's grow together. Drop me a DM! 🤝
Edinson Carranza retweeted
Hitting 1M impressions on X feels harder than ever.
What used to work:
- threads
- hot takes
- posting 10x a day
What seems to work now:
- replies that start real conversations
- posts people actually save
- staying in one niche
The algorithm didn't kill reach.
It killed lazy reach.
Agree or not?
Edinson Carranza retweeted
🚨 NOVEMBER IS solana:8T3suJtKUGrWRytVNKe7RLV81AumvmBPQfEkyeHtpump MONTH.
The world's most powerful AI models are coming to VIAINTI at ridiculously low prices.
But here's the real deal:
solana:8T3suJtKUGrWRytVNKe7RLV81AumvmBPQfEkyeHtpump holders get 6 MONTHS of FREE ACCESS.
No expensive subscriptions.
No overpriced AI credits.
Just powerful AI, accessible to everyone.
We're not building another AI wrapper.
We're changing how people access AI.
November changes everything.
Edinson Carranza retweeted
Creators: be honest.
Is it harder to get reach on X now, or did our content just get worse?
What changed for you in the last 3 months?
Edinson Carranza retweeted
AI made me 10x faster at writing code.
It made me 0x faster at knowing what's worth building.
That part is still on me.
What's the last thing you built fast and regretted building at all?
Edinson Carranza retweeted
Replying to @blknoiz06
Agree the token is hard to value. That's the problem, not a feature.
If apps start capturing clear revenue, L1 tokens have to justify themselves as money.
What metric would you give a TradFi analyst for SOL today: fees, staked supply, or stablecoin float?
Edinson Carranza retweeted
🚨 Is the AI bubble starting to burst?
China's STAR 50 ETF is down 7.89% YTD, reflecting pressure on some of the country's leading technology companies.
China's tech sector is already feeling the pain. Could Wall Street be next?
The US is investing hundreds of billions into AI infrastructure, driven by expectations of massive future profits.
But there's a growing question: Will AI revenues justify these valuations?
China's decline doesn't prove the AI bubble has burst. But it's a warning that technological innovation doesn't guarantee investment returns.
AI is real. The technology is revolutionary. But that doesn't mean every AI company is worth its valuation.
What happens if investors stop paying for promises and start demanding profits?
The next major market correction could begin with AI.
Edinson Carranza retweeted
Genie 4 is out as a research preview.
You steer a live world with text: weather, camera, new objects. With sound.
Per Google, Gemini is the "brain" of the simulation.
If prompts can steer a world, which part of game dev gets automated first: art, level design or QA?
Edinson Carranza retweeted
🚨 BITCOIN: THE REAL TEST IS COMING.
$BTC is facing uncertainty ahead of the US midterm elections and the Fed's next moves.
My outlook:
📉 Bearish scenario: $74K
📈 Recovery target: $95K
🔥 Bullish scenario: A strong year-end rally.
Historically, Bitcoin has delivered significant gains in the years following US midterm elections.
But history doesn't guarantee another rally.
The real question isn't whether Bitcoin can reach $95K.
It's whether buyers will return when fear reaches its peak.
Are you accumulating or waiting for lower prices?
Edinson Carranza retweeted
Replying to @GoogleDeepMind
Interesting that Gemini plans events inside the sim instead of just rendering frames.
The real test is how long a scene stays coherent before it drifts. Anyone seen how long a session runs?
Edinson Carranza retweeted
Cursor just shipped an Agent SDK.
Agents as a folder.
Prompts in Markdown.
Tools in TypeScript.
Built-in evals.
Good. Agents finally get code review.
But most evals check one thing:
did it finish the task?
The eval I'd write first:
did it refuse the task it was never supposed to do?
What's actually in your agent eval suite?
Edinson Carranza retweeted
Bubblemaps alleges a top-3 Fomo trader with 200K+ followers is tied to bundled memecoins.
Their numbers:
• public wallet: -$70K
• connected wallets: +$270K
His reply: he backs teams he trusts that don't rug. Alleged, not proven.
Before you copy a KOL call, check:
1. fresh wallets buying at the same time
2. who funded the dev
3. who sold into the pump
If his public wallet is -$70K and linked wallets are +$270K, whose PnL do you trust?
Edinson Carranza retweeted
Claude Code now has an env var called
CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS=1
Plain skip-permissions still blocks obvious stuff like rm -rf ~. This removes that.
Before you give an agent that much rope:
• run it in a container
• no prod creds in env
• clean git state first
What can your agent never touch, even with permissions off?






