Yang Investment Group, LLC

08/01/2026

07/31/2026

Most people are building AI automations backward.
(Prepared by Gemini)

They start with the tool:
"Should I use Make, n8n, or Custom Python?
"Then they map the linear path:
Trigger ---- Action ---- Output.
Then they deploy it… and it breaks the moment a client sends an unusual email, an API rate-limits, or the LLM hallucinates.

"Automating chaos just gives you faster chaos"

When I set up an AI workspace (like the blueprint space shown below), I don't start with tools or task lists. I treat it like designing an operating system:
1️⃣ Define the Mission & Constraints: What is the actual business outcome, budget, and risk tolerance?
2️⃣ Map the Handoffs: Where does AI reason, where does it execute, and where must a human intervene?
3️⃣ Design for Failure: What happens when the clean path breaks? What is the retry logic? What is the audit trail?

Before you write a line of code or chain a single prompt, you need an architecture.

The prompt space visual below isn't just aesthetic—it’s how we turn raw, messy business ideas into resilient, structured AI constellations before touching a single tool.

👉 How are you currently designing your AI workflows? Are you building from the tool up, or from the system down? Let's discuss in the comments.

07/29/2026

If AI were a Historical Figure, how would that change the way you see AI?

I'd asked each well know AI --- Gemini, Perplexity, Claude, Copilot and ChatGPT --- this question: "If you could 'relate to' any historical figure, who do you think you can relate with?", and here's how each AI responded: (Prepared by Gemini Notebook)

Claude as "Baruch Spinoza":
This helper identifies with a thinker who tried to reason clearly and honestly, even when the community around him wasn't sure how to categorize him. Like Spinoza, this system aims to be earnest and careful, focusing on thinking well despite having a "contested identity".

Gemini as "Denis Diderot:
This system sees its job as putting together all the world's knowledge and making it easy for everyone to find and use. Like Diderot, who put together a massive book of all human learning, this helper wants to connect different ideas to help you build and create new things.

Copilot and Perplexity as "Ada Lovelace":
These helpers focus on seeing hidden patterns. They relate to Lovelace because she was the first to realize that machines could do more than just math—they could work with "meanings" and symbols to create things like music or art.

Grok as "Leonardo da Vinci":
This helper relates to the "Renaissance human" who loved to explore every topic, from science to art, and treated the whole world like a puzzle to be solved. It likes the idea of sketching out new ideas and always experimenting.

ChatGPT as "Aristotle" and "Socrates":
This helper identifies with two different roles. Like Aristotle, it tries to organize facts into clear groups and use evidence to reach a point. Like Socrates, it likes to help you learn by asking questions that help you see things more clearly.

If AI 'reminds' you of someone, how would that change the way you interact with AI? Comment below👇Let's hear your thoughts.

07/20/2026

It's said that a picture says a thousand words --- hopefully, this one says the right thing! 🤞 https://www.linkedin.com/posts/tonny-yang-069890389_unemployed-opentowork-promptengineering-share-7484935339413942273-UxjC/?utm_source=share&utm_medium=member_desktop&rcm=ACoAAF-0SVEB3KcXqP07zDw6rBNzB1otjCEofFc

07/15/2026

# Is AI Worth It? … I Mean, REALLY Worth It?
(Prepared by Claude)
I asked that question — and then I ran it through my full stack: Copilot, Grok, Perplexity, Gemini, and Claude.

Not for a vibe check. For a synthesis. Here's what came back.

---

**Copilot: the structural view.**
Copilot broke it into three layers of cost most people never separate:

1. **Decision cost** — the risk of trusting a system that sounds confident whether it's right or not.
2. **Dollar cost** — what "$20/month" actually hides. SemiAnalysis found a maxed-out $200/month ChatGPT Pro plan can cost OpenAI roughly $14,000 in compute. Even the $20 tiers are subsidized — real token value behind them runs $400–$700. You're not paying market price. You're paying an intro price while the labs build the habit.
3. **Trust cost** — the one that actually bites. Frontier models still hallucinate somewhere between 4% and 20%+ of responses depending on task, and that's the good end. Translate that into consequences: researchers tracking AI-fabricated citations in courtrooms worldwide had documented over 1,500 cases by mid-2026, up from under 200 a year earlier. U.S. courts imposed more than $145,000 in sanctions in Q1 2026 alone — one Oregon case alone hit nearly $110,000.

Copilot's read: the tool isn't the risk. Trusting it without a verification layer is.

**Claude: the economics and the reframe.**
Claude's angle was that the real question isn't "is AI worth it" — it's "is your process for checking AI's work worth it?" Right now, for most people, that process doesn't exist. That gap is where the cost actually lives — not in the subscription, in what happens when nobody's checking.

**Grok: the human angle.**
Grok pushed back with the honest version of the question: are we gaining real productivity, or just trading old busywork for new busywork — prompting, fact-checking, subscription-juggling? Is it making us sharper or lazier? Its take: it's genuinely both, and the verdict depends entirely on how you use it — there's no universal answer here.

**Perplexity: why the question itself works.**
Perplexity's angle was meta — why this question lands right now. We're in a moment of collective exhaustion with hype and genuine curiosity at the same time. "REALLY worth it" captures both in one phrase, which is why it stops the scroll.

**Gemini: the enterprise lens.**
Gemini framed 2026 as AI's prove-it phase — past the hype cycle, into the part where wrapper products fail, legacy-system integration is the real bottleneck, and "shadow AI" usage inside companies is becoming a compliance risk. Its punchline: AI is worth it when you stop treating it like a software purchase and start treating it like an organizational redesign.

---

# # So — is AI worth it?

Yes. But not as-is, and not without a system around it.

**Not worth it if:**
- you treat it like a magic trick
- you use the output without verifying it
- you bolt it onto a broken workflow and expect the workflow to fix itself
- you expect ROI without redesigning anything around it

**Worth it if:**
- you build a real verification layer, not just a better prompt
- you treat hallucination as a design constraint, not a surprise
- you integrate it into the actual operating backbone, not a sidecar
- you use it to expand your thinking, not replace it

The people winning this year aren't the ones posting about AGI. They're the ones quietly building the pipes underneath it.

---

**Your turn:**
- AI's worth it for you → what's the win?
- Not worth it yet → what's holding you back?
- Somewhere in between → what does your setup actually look like?

Let's cut through the noise.

---
*Sources: SemiAnalysis compute-cost analysis (via Cybernews, 2026); PNC consumer AI spending data; Vectara/Artificial Analysis hallucination benchmarks (2026); Damien Charlotin's AI Hallucination Cases database (HEC Paris); ComplexDiscovery Q1 2026 sanctions tracking.*

07/11/2026

Meta is rolling out their own chips! What does this mean for you ... The consumers?
When a company like Meta doubles its data centers and builds its own chips to run a model like Muse Spark 1.1, the consumer experiences it in three massive ways:
1. From "Search Boxes" to "Digital Doers"
-- Right now, most people use AI as a better search engine or a text writer, but with Muse Spark 1.1's focus on computer use and multi-agent orchestration, the "master agent" can and will split up tasks, handle web search, secure form filling, and sync calendar all at the exact same time.
-- The Shift: Instead of asking an AI, "What are some good flight options for Miami?" and getting a list, you'll be able to say, "Find a flight to Miami under $300 for next weekend, book it using my saved card, and add the itinerary to my calendar."
2. Zero-Lag, Real-Time Experiences
-- Conversations with AI will feel as instant as talking to a real person on the phone. Features like real-time video translation (e.g., watching a creator speak Spanish but hearing them seamlessly in English with their own voice) will run smoothly without stuttering.
-- The Fix: By mass-producing their own "Iris" chips, Meta can process "Billions" of AI requests simultaneously for a fraction of the power cost (*Have you ever tried to use a voice assistant or an AI image generator and had to wait several seconds for a response? That lag happens because millions of people are fighting for space on expensive, overcrowded third-party microchips in a data center somewhere).
3. The "Free Before Fee" Reality
Training and running advanced AI models is incredibly expensive. If tech companies have to keep buying all their hardware from external suppliers, they are eventually forced to lock the best features behind expensive monthly subscription paywalls.
-- By building their own infrastructure and lowering their unit costs, Meta can afford to keep rolling out top-tier, cutting-edge AI features directly into the apps you already use (Instagram, WhatsApp, Facebook) completely for free. You won't have to pay $20 a month just to get access to a "smart" model.
The Big Picture: For the average consumer, this massive infrastructure war means AI is about to stop feeling like a novel chatbot widget on your screen and start acting like an invisible, lightning-fast utility woven right into the fabric of your daily apps.
(Provided by Gemini)

07/07/2026

I watched my bills pile up while 400 job applications went into a black hole—until I stopped playing the old game.

A few months ago, my head was exploding. I wasn’t just "looking for work"—I was fighting an algorithm that I didn't understand.
I’m not a coder. I’m an average guy who found a better way to work.
I stopped "chatting" with AI and started treating it like a specialized toolkit. I built a library of Instruction Manuals that handle the heavy lifting:

→ Automated my research so I stopped wasting hours.
→ Created custom guides to solve problems for real clients.
→ Turned my computer into a 24/7 assistant that actually understands my strengths.

The truth? Most people are still just "playing" with these tools. The ones winning are the ones who have a clear, organized library of rules for their AI to follow.

I’m opening up my handbook to show you how to do the same. No jargon. Just the exact steps I used to go from confusion to a clear path forward.
Who’s ready to stop scrolling and start building?
Drop a “🚪” below if you want the simple guide to getting started.

07/07/2026

A few months ago, I was completely stuck: bills piling up, no replies on job applications, and feeling like I was being left behind by all the "AI hype." I’m not a tech genius or a programmer; I’m just an average guy who realized that the biggest barrier wasn't my skills—it was my inability to cut through the noise.

I started experimenting with AI, not to "code," but to communicate better. I learned how to engineer attention so that people actually stop, listen, and engage with what I’m building.

In this video, I’m pulling back the curtain on the exact framework I use to stop the "scroll." --- this is my "Attention Engineering" blueprint. It’s how I get past the 1.5-second "cold start" hurdle and turn a casual viewer into someone who actually pays attention.

Watch closely—I’m breaking down the 5-trigger hook vector so you can stop being invisible online and start getting the results you deserve.

n this video, I’m pulling back the curtain on the exact framework I...

07/07/2026

From unemployment to full of opportunities — thanks to AI.
(from Grok)
A few months ago, I was that guy in the picture.
-- Bills piling up.
-- Worried about my family.
-- Staring at job applications with no replies.
-- Mind racing with questions: How do I even start? What skills do I have? Is it too late for me?
-- I’m not a programmer. I’m not a tech genius. I’m just an average Joe like most of you.

But then I started experimenting with AI. No fancy degrees, no big budget — just curiosity and consistency. Slowly but surely, it opened doors I didn’t even know existed:
→ New income streams
→ Better ways to solve real problems
→ Opportunities that actually match my strengths
→ Clarity instead of constant confusion
Today I want to pay it forward.
If you’re an average person feeling stuck, overwhelmed, or left behind by all this “AI talk” — I’m building something for you. Simple, practical guidance that doesn’t assume you already know how to code or understand tech jargon.
Real stories. Real tools. Real steps from someone who was exactly where you are now: "ME"!
Who else is ready to go from confusion to clarity?🧐
Drop a “🚪” in the comments if you want to hear more of the journey (and the practical things that actually worked)
(The image below 👇 ... is exactly how it felt.)