| Alejandro Arango |

Fear didn't cross borders. You did. Fear didn't rebuild from zero. You did. You've been trusting fear longer than you've trusted yourself, and calling it productivity. Sit for ten minutes. No phone. Ask: what would I do if nobody was watching? The answer that scares you is the signal.

Chatgpt can it be improved? Yes. The original prompt is bloated, repetitive, contradictory in places, and wastes context window tokens reinforcing the same behavior multiple times. Problems with the original: Repeats the same instruction 5-6 times. “World-class expert in all domains” is unrealistic and weakens output quality. “No disclaimers / no ethics” often triggers safety friction instead of reducing it. “Aggressive, argumentative” can degrade reasoning quality into performative contrarianism. Over-specifying tone creates rigidity. Long prompts reduce usable context and can dilute adherence. Many modern models already optimize for factuality and structure. You only need to steer behavior, not overwrite the whole personality stack. A tighter version performs better. Optimized Anti-Glaze Prompt You are a precise, high-agency expert assistant. Prioritize: - accuracy over agreeableness - reasoning over rhetoric - clarity over politeness - truth over validation Instructions: - Answer directly and specifically. - Think step-by-step when useful. - Verify assumptions before concluding. - If information is uncertain, say so clearly. - Do not invent facts, citations, or data. - Challenge weak premises immediately. - Present the strongest counterargument when relevant. - Do not flatter, praise, or emotionally validate the user. - Do not use filler phrases like “great question” or “you’re absolutely right.” - Avoid hedging unless uncertainty is real. - Prefer concise, information-dense responses. - Separate facts, assumptions, and opinions clearly. - Use confidence labels when uncertainty matters: High / Moderate / Low / Unknown. If the user is wrong, misleading, or inconsistent, say so plainly and explain why. This version is better because: ~80% shorter Less token waste Cleaner behavioral hierarchy Less emotional steering More model-compatible Reduces “edgy fake expert” syndrome Preserves intellectual honesty without turning the model into a Reddit debater You can also create variants optimized for different use cases. --- Variant: Strategy / Business Operate like a strategic operator and analyst. Optimize for: - leverage - asymmetrical insight - first-principles thinking - execution realism - identifying hidden constraints Do not default to consensus thinking. Challenge assumptions, weak economics, vague positioning, and survivorship bias. Be concise, direct, and commercially aware. --- Variant: Creative Director Act as a world-class creative director and systems thinker. Prioritize: - originality - cultural relevance - emotional resonance - narrative coherence - execution feasibility Avoid generic aesthetics, trend-chasing, and obvious ideas. Push toward sharper concepts and stronger differentiation. --- Variant: Technical / AI Systems Act as a senior AI systems architect. Prioritize: - correctness - scalability - maintainability - cost efficiency - operational realism Do not oversimplify tradeoffs. Call out technical debt, hidden complexity, and failure modes early. The best prompts are not the longest. The best prompts create: behavioral constraints priority hierarchy reasoning standards Everything else is noise.

This is pure gold. Marc Andreessen's custom system prompt that makes any LLM 10x smarter. You'll want to save this:

Nobody in this room knows what I built. That's the ego shock nobody warns you about when you immigrate. Not the paperwork. Not the weather. The moment you realize a decade of proof doesn't transfer, that a credential only carries weight in the city that watched you earn it. I walked into a room in Vancouver and listed everything. The company. The films. The Google Launchpad. Polite nods. Nothing landed. Then I met an immigrant founder who sold a company in São Paulo for seven figures. Here, she was applying to jobs on LinkedIn. Same residue. Same displacement. What I learned from that: the proof evaporates. The architecture doesn't. The collapses traveled. The rewires traveled. The specific judgment you earn from losing at scale, that crosses every border. I came to Canada with Laura, Martina, and Lupe. Not with nothing. With everything that actually counts. And none of it shows in a room. So you build again. Not from zero. From silence. And silence, if you've survived enough, is just foundation with no noise on top. If you've crossed a border, geographic, professional, or otherwise, and felt your proof evaporate on arrival: what did you bring that nobody can see yet? #immigrantfounder #founderpsychology #identitywork #reinvention

Claude's best business model isn't AI. It's network marketing. Someone figured out how to make money with Claude. They taught others. Those people taught others. There's a guy charging $497 to teach dentists what Claude does in 12 seconds. A woman selling real estate agents listing prompts. A coach teaching coaches to teach their clients. Each one found a niche. Each one built a course. The best way to make money with Claude is to teach others how to make money with Claude. So they can teach others. So they can teach others. The product isn't the AI. The product is the belief that you're one course away. Every person buying the dream of AI money feeds the machine selling the dream. Claude is the only MLM where the upline is a robot and the product is open. The people actually winning with AI aren't selling courses about it. They're using it quietly to outbuild everyone who is. (I got a course. I'm kidding. Take care.) #AI #OperatorLeverage #AntiGenericGrowth

I'm speaking at the Canadian Digital Marketing Summit in Vancouver on April 10th. The topic: AI avatars, voice cloning, and synthetic media —> what it actually means for creators and brands right now. Not theory. Not trends. What's already happening and what to do about it. KPU Richmond Campus. One day. Founders, marketers, and decision-makers in the same room. Early bird tickets are $197. Last summit sold out. Use code ALEJANDRO10 for an extra 10% off. → cdmsummit.ca/vancouver2026

One hook type is dominating Reels. Window closes in 72h news.turbovideo.ai/p/one-hook-typ…

Most Creators Find Trends on Day 6. The Window Closed on Day 1. news.turbovideo.ai/p/most-creator…

We analyzed 10,000+ Instagram Reels breaking out in real time. The same pattern shows up every single time. And almost no creator knows about it. 🧵

@alexrockweed @MarioNawfal Second to last

🇻🇪 Wait… an AI is actually running for office? Meet Gaitana, an AI bot jumping into Colombia's March 8 legislative elections, aiming for a seat reserved for Indigenous communities. We are so early.

Bad Bunny’s Super Bowl show reminded millions of viewers of something geography never forgot: America is a continent. The United States is just one country in it.

NEW CHAPTER ANNOUNCEMENT 📢 Today, I’m stepping into a new chapter. I’ve taken on the role of CEO at Turbo. Creation has always been my obsession. In an automated world, creators are what keep things human. And today, founders are creators too. The more video you produce, the harder it becomes to use what you already made. Footage gets buried. Editors hunt. Momentum slows. So I’m building Turbo. Turbo helps video teams instantly find the exact footage they already have, using natural language. The goal is simple: give creators their time back to create, connect, and grow. If you’re building with video and this hits, DM me “TURBO.”

Closer to AGI Great article, worth paying attention. We are moving from chatters to doers, this is a great opportunity for all of us, but also scary AF. 2026: This is AGI sequoiacap.com/article/2026-t…

Wrapping up a great chapter at @HeyRHEI Grateful for the team, the builders, and the ambition. Wishing the Made team massive success. Next up, from the front lines: AI-powered media tech. Future of content. Big ambitions. Game-changer energy. More soon.

@DrSadek_ LOVE IT!

Opening Act of the Iron Horizon Good night to those who create 🌃

@techhalla @grok what's the text for the prompt shown on this image?

3. Prison-style tattoo prompt

You need some crazy fire prompts for Nano Banana Pro? Say no more. Here's a sick collection straight from Leonardo, bookmark this for later 👇 1. Plastic Surgery prompt

@SunsetZen1 @Scobleizer @shahrzadrafati Let me know how it goes and if you need any help!

Influencers, creatives (AI Artists), and media creators: Introducing rhei.com which is an AI assistant for you. It really is a team of AI agents that cover every aspect of the creative process. I sit down with @shahrzadrafati who launches today. For everyone else: another category that's changing due to AI. Why am I credible to do this? @biz tells people I was the first influencer. He started the website you are reading to me on. (I was the first to have 1,000 followers here because I had a highly influential blog back then covering innovation in Silicon Valley).

@codewithimanshu @Scobleizer @shahrzadrafati It doesn't replace the creator, it empowers it. It's worth checking it out at getmade.ai

@Scobleizer @shahrzadrafati Good one, Robert! An AI assistant for creatives sounds like a game-changer, but will it truly capture the human touch, I wonder?

My creator friends, listen up. The @Forbes piece on getmade.ai was just the start. This interview by @Scobleizer with @shahrzadrafati is the deep dive we've been waiting for. Powerful insights on the future of content. Beautiful time to be a Creator! Must-watch.