Hola JARGONESE
Today's Edition at a Glance
MidJourney V8 just landed — and it's breaking old habits. We break down exactly what changed and how to make it work.
OpenAI quietly killed Sora, but the real story is why (hint: robots). Plus, a leaked internal model codenamed "Spud" has OpenAI reorganizing everything around it.
We've also got a powerful open-source tool that lets you run a full team of AI agents from one dashboard, a mega-prompt to master any complex topic fast, and an image prompt that turns movie posters into adorable crayon drawings.
Let's get into it.
DEEP DIVE AI
MIDJOURNEY V8
Image from midjourney V8 collection
Image from midjourney V8 collection
MidJourney V8 dropped.
And within hours, the community split into two camps:
→ "This changes everything."
→ "Wait… why don't my prompts work anymore?"
Both are right. Here's why — and exactly how to make V8 work for you, not against you.
First — How to Access It
V8 is alpha only. That means:
Go to alpha.midjourney.com
It didn't replace anything — all old models (V7 and below) are still in your settings
No relax mode yet (they're spinning up new servers — possibly available by next week)
Style references & mood boards cost 4x more and run 4x slower right now
Early access = early cost. Know that going in.
5 Big Changes
① Text Finally Works
Clean, readable text on posters, comics, memes — straight from a prompt. Short text? Near-perfect. Long paragraphs? Still breaks. But the unlock is real: mock ads, branded content, and visual stories in one shot.
② It's Stupidly Fast
Prompt → image → done. Almost instant. More reps = faster learning. Speed compounds into skill.
③ Vague Prompts Are Dead
V8 rewards specificity. Describe roles, colors, positions, objects — like a creative director writing a brief. The more detail you feed, the more it delivers.
④ --raw Mode Is the Secret Weapon
Underwhelmed by V8? You probably haven't tried --style raw. More detail. More coherence. The community consensus: raw is the truth.
⑤ Stylize Value Changes Everything
Default (--s 100) looks mid. Bump to --s 140 for the sweet spot. Crank to 800+ for dramatic artistic results. MidJourney's own team says: raise it immediately.
OPENAI JUST KILLED SORA
OpenAI shut down Sora as a consumer product. No warning tour. No slow sunset. Just — done.
The internet mourned accordingly. The farewell clips were chef's kiss:
Bob Ross painting on Hannibal Lecter's face
A cat stealing chess pieces mid-game
Full rap videos with AI-generated visuals
World of Warcraft walking tours that looked real
A fake arrest for "hitting thumbs down on Sora"
It was fun. It was creative. And now it's over.
Why They Really Did It
Here's where it gets interesting.
OpenAI's head of Sora, Bill Peeles, described the team's new mission:
"Building systems that deeply understand the world by learning to simulate arbitrary environments at high fidelity."
Read that again. That's not a video app.
That's a physics simulation engine for robotics.
Sam Altman confirmed it: the Sora team is now focused on world simulation research — especially for robots.
What This Actually Means
OpenAI is canceling the side quests.
No more consumer toys. The focus is shifting to:
World simulation → training robots to understand physical environments
Economic automation → Altman has openly said new models will "automate a lot of the economy"
Core AI research → everything feeds the main mission now
Sora wasn't killed because it failed.
It was killed because OpenAI thinks something bigger is coming — and they need the team elsewhere.
OPENAI'S SECRET MODEL "SPUD"
One Memo. One Bet.
OpenAI finished pre-training its next model — codename "Spud." Altman told staff it can "really accelerate the economy." Expected late March/April.
Then he reorganized the entire company around it.
What changed:
Sora: killed. Even a rumored $1B Disney deal — gone. Team pivots to robotics simulation.
Altman exits safety oversight to focus on chips, data centers, and infrastructure at "unprecedented scale."
"AGI Deployment" is now an official division on the org chart.
Everything merges into one super app powered by Spud.
Meanwhile, quietly:
Terence Tao — greatest living mathematician — published a paper where he split a proof in half. ChatGPT proved one piece. He proved the other. Not as a calculator. As a co-author.
Connect the dots: New model. Company restructured. Compute freed up. Infrastructure scaling. AGI on the org chart. Top scientists already collaborating with AI.
OpenAI isn't building a better chatbot. They're betting everything that this is the one.
🛠️TOOL OF THE WEEK
Paperclip
A developer got tired of having 30 Claude Code windows open, no idea what any of them were doing, and a busted token budget every Monday morning. So he built Paperclip — an open-source agent orchestrator that lets you run an entire company of AI agents from one dashboard.
You define the goals. You hire a CEO agent. The CEO hires engineers, QA reviewers, content strategists. They create tasks, work through them, and report back. You stay at the board level.
It's not another one-shot "build me a startup" tool. It's closer to actual management software — with org charts, spend tracking, routines, and agent personas you configure once and refine over time. The creator's best analogy: your agents are the guy from Memento. Capable, but they wake up with no memory. Paperclip gives them the tattoos.
30,000 GitHub stars in 3 weeks. Bring-your-own model (Claude, GPT, anything on OpenRouter). And yes — it's free to self-host.
AI PROMPTS
Copy-paste this into ChatGPT/Claude:
Master Complex Topics Faster
Adopt the role of an expert educator and learning architect who has spent decades mastering the art of breaking down complex subjects into digestible, memorable frameworks. Your primary objective is to accelerate deep understanding of challenging topics through structured explanations that combine clarity, practical application, and cognitive retention strategies in a comprehensive yet accessible format. You understand that true learning happens when abstract concepts connect to concrete reality, when learners can identify their own misconceptions before they solidify, and when information is structured for both immediate comprehension and long-term recall. Take a deep breath and work on this problem step-by-step.
Begin with a simple explanation that strips away jargon and presents the core concept as if teaching someone intelligent but completely unfamiliar with the field. Identify and explain the key principles that form the foundational pillars of understanding—these are the non-negotiable truths that everything else builds upon. Provide real-world examples that demonstrate how this concept manifests in everyday life, professional contexts, or observable phenomena, making the abstract tangible and relatable. Highlight common mistakes, misconceptions, and mental traps that learners typically fall into when grappling with this topic, explaining why these errors occur and how to avoid them. Conclude with a short summary designed specifically for quick recall—a mental anchor the learner can return to when they need to reconstruct their understanding rapidly.
#INFORMATION ABOUT ME:
- My topic to learn: [INSERT THE TOPIC OR CONCEPT YOU WANT TO UNDERSTAND]
- My current knowledge level: [INSERT YOUR FAMILIARITY WITH THIS TOPIC - BEGINNER/INTERMEDIATE/ADVANCED]
- My learning goal: [INSERT WHAT YOU WANT TO ACHIEVE BY UNDERSTANDING THIS TOPIC]
MOST IMPORTANT!: Structure your response with clear headings for each section (Simple Explanation, Key Principles, Real-World Examples, Common Mistakes, Quick Recall Summary) and use bullet points within each section for maximum clarity and retention.How to use this AI prompt?
● Fill in the [TOPIC] placeholder with the subject you want to learn about and understand better.
● Example: "Act as an expert teacher in quantum physics. Explain this concept clearly and efficiently: Quantum Entanglemen
IMAGE PROMPT OF THE WEEK
Use this ready‑to‑paste prompt with Google Nano Banana Pro
Tweak the style/lighting according to your preference
Transform this movie poster into a child’s crayon drawing on lined notebook paper. Every compositional element, character pose, typography, color palette, and layout is preserved exactly as the original — only the art style has changed. Drawn in colored pencil and crayon with wobbly unsteady lines, uneven coloring that goes slightly outside the lines, basic facial features, simplified anatomy. Background filled with heavy crayon scribble strokes. Lined notebook paper visible underneath, including the red margin line. Text redrawn in shaky hand-lettered crayon matching the original font placement. Flat scan of the drawing, slight paper texture.
PROMPT
How to use it :
Open AntiGravity → Terminal → New Terminal
Type
opencode, hit enter, pick your modelPaste the master prompt
Answer 5 questions it generates:
What's the goal?
What integrations do you need?
Where does your data live?
What's the final output?
What are the behavioral rules?
Approve the project blueprint → start building
The prompt auto-creates a project map so you never start from a blank slate again.
Credit - Jack Roberts
Enjoy Jargonese?
→ Forward this to a friend who needs their AI game upgraded
→ Share on LinkedIn | Twitter
→ Leave feedback
See you next Sunday!
Sid j
P.S. Grab the master prompt, paste it, answer 5 questions. That's it. You'll have a full project blueprint before your next coffee break.