AI DUNN Right Weekly - Issue #12

Practical AI insights for business growth

Hey AI Innovators!

Welcome back to AI DUNN Right Weekly! This week the AI world went code red (literally), music labels stopped suing and started partnering, and OpenAI discovered building the future costs way more than anyone thought. Plus some wild tools that actually solve real problems.

Here's what matters for your business.

Read time: 4 minutes

---

This Week's Game Changer

OpenAI Hit Code Red - And It's Good News for You

What happened: OpenAI declared "code red" and paused everything (shopping features, health agents, even ads) to fix ChatGPT's core issues. They're rushing out a new reasoning model next week that internal tests say beats Google's Gemini 3. The leaked Sam Altman memo shows they're going all-in on speed, reasoning accuracy, and actually answering questions right. (Source: Financial Times, What's Up in AI)

Why it matters: For once, a tech company stopped chasing shiny new features and decided to make the core product actually work. If you've been frustrated with ChatGPT giving you slow, wrong, or confusing answers, this is the fix you've been waiting for. They're redirecting compute and staff away from monetization experiments to make the thing you actually use every day better.

Here's what I think: About time. ChatGPT has been getting bloated with features nobody asked for while the basics keep breaking. I've lost count of how many times it's hallucinated sources or taken forever to spit out a simple answer. Hitting code red because Google's Gemini is eating your lunch? That's the kind of pressure that actually ships improvements instead of promises.

The fact that they're pausing revenue-generating products to fix core quality tells you they're actually worried. That's when companies ship their best work.

Business impact:

• Faster responses mean less waiting around for basic queries • Better reasoning means fewer garbage outputs you have to redo • Visual quality upgrades make it useful for design and creative work • This forces Google to keep improving Gemini, which means we all win

What to do: Expect a big ChatGPT upgrade next week. Test it against your current workflows and see if the improvements actually landed. If you've been considering switching to Gemini, hold off until you see what OpenAI ships.

---

What's New This Week

Warner Music Just Made Peace with AI (After Suing It)

What happened: Warner Music Group partnered with Suno AI - the same company they were suing six months ago for copyright violations. Now they're building licensed AI music tools that let artists generate songs using their own voice, style, and catalogs. Artists get full opt-in control, meaning nobody's getting ripped off. (Source: BBC, What's Up in AI)

Here's what I think: This is what happens when you realize the tidal wave is too big to fight. Warner looked at Suno users generating an entire Spotify catalog every two weeks and said "we either partner or get left behind." Smart move.

Six AI artists have already charted on Billboard. The music industry learned from what happened to newspapers and taxis - fight new tech and you die, partner and you might survive. Artists keeping control over their likeness and getting paid for AI-generated versions of their work? That's the model every creative industry should be watching.

Who this matters for: Content creators, marketing teams using music, anyone in creative industries wondering how AI licensing will actually work. This is the template.

---

Tool of the Week

Oboe - The $16M AI Learning Platform That Actually Teaches

What it does: Oboe (just raised $16M from a16z) builds custom curriculums for whatever you want to learn, then explains complex concepts with visuals that make sense. No more bouncing between random YouTube videos hoping something clicks. You tell it "learn Python for data analysis" and it maps out exactly what you need, with diagrams and explanations that stick. (Source: TechCrunch, What's Up in AI)

Why you care: Training your team costs a fortune. Traditional courses are too generic, hiring trainers costs $50K+, and sending people to YouTube wastes weeks. Oboe creates structured learning paths with the exact skills your team needs, nothing more.

One fintech startup used it to teach their engineers blockchain fundamentals. Result? They shipped the product three weeks early because people learned exactly what mattered instead of wading through bloated courses.

Here's what I think: Most online learning is garbage. You get either 200-hour comprehensive courses that teach you stuff you'll never use, or random 10-minute videos that skip the important parts. Oboe fixes this by asking "what do you actually need to do?" and teaching you that.

The visual explanations are what sold me. One developer said Oboe's memory management diagrams made Rust finally click after text tutorials failed for months. That's the difference between theory and actually understanding something.

Who should use it:

• Teams that need upskilling without hiring expensive trainers • Founders learning technical skills fast • Managers who need their people trained on new tools without losing weeks of productivity

Try it: Sign up at oboe.com (currently in beta with waitlist access)

---

Quick Hits

Three More Things You Should Know:

1. OpenAI Won't Make Money Until 2030 A financial analysis estimates OpenAI needs $207 billion to fund its AI ambitions and won't be cash-flow positive before 2030. They're burning $8.65 billion in nine months just on infrastructure and inference costs. Revenue's hitting $20B annualized, but compute partnerships with Azure, CoreWeave, AWS, Google, and Oracle still can't meet demand. (Source: Medium, Yahoo Finance)

Here's what I think: This is the AI paradox nobody wants to talk about. The companies building the future are hemorrhaging cash faster than any sector in tech history. Models double compute needs every 3-6 months, which means even with record revenue, OpenAI can't keep up with its own growth. This is why every major AI company is desperately raising billions - they're in a race where second place means irrelevance, but winning costs more than anyone imagined.

2. ChatGPT Voice Mode Goes Native OpenAI merged Voice Mode into the main ChatGPT interface. No more separate voice experience, just switch between text and voice in one window. Works across GPT-5.1 Instant and Thinking, supports live conversation, real-time interruptions, and emotional tone. (Source: TechCrunch)

Here's what I think: Voice is becoming the default interface and OpenAI finally admitted it. Folding it into the main UI means they're done treating it like an experiment. If you've been typing when you could be talking, this makes it way easier to switch. Especially useful when you're hands-free or need to brainstorm out loud.

3. Trump's Genesis Mission Wants to Double Science in 10 Years President Trump signed an executive order launching the Genesis Mission - a Manhattan Project-style AI program to double U.S. research productivity within 10 years. DOE will build an AI platform connecting supercomputers, robotic labs, and scientific datasets. Focus areas: fusion, quantum, advanced materials, drug discovery. (Source: White House, Le Monde)

Here's what I think: This is $200 billion worth of scientific data nobody could access before, now paired with the world's biggest public supercomputing network. If it works, this shrinks decade-long research cycles to months. If it doesn't, it's a spectacular waste of taxpayer money. (Nothing new there....) Either way, every private AI company just got a government-funded competitor with unlimited resources. Buckle up.

---

Prompt of the Week

The "Don't Skip Steps" Framework for Better AI Reasoning

Use this when you need ChatGPT to actually think through complex problems instead of jumping to conclusions:

I need help analyzing [DECISION/PROBLEM].

Before giving me an answer, walk through this framework:

1. Clarify the core question - What am I actually trying to solve?
2. Identify the constraints - What limits my options? (budget, timeline, resources, etc.)
3. List the stakeholders - Who's affected by this decision?
4. Map the trade-offs - What do I gain vs. lose with each option?
5. Consider second-order effects - What happens after the immediate result?
6. Provide your recommendation with reasoning

Work through each step explicitly. Don't skip ahead.

Why it works: This forces AI to show its work instead of giving you the first answer it thinks of. You catch bad assumptions early, and the step-by-step breakdown makes it obvious when the logic falls apart.

Example use: "I need help to analyze whether to hire a full-time designer or use freelancers for our Q1 campaign." The prompt walks through budget limits, stakeholder needs (marketing team, clients, brand), trade-offs (quality vs flexibility), and what happens after you make the choice.

Here's what I think: Most people treat AI like a magic 8-ball - shake it and hope. This prompt forces it to actually work through the problem like a consultant would. The "don't skip ahead" part matters because AI loves conclusions without showing work. Make it earn the answer.

That's it for Issue #12!

Next week: more AI news, fewer features that don't matter, and my actual opinions on what's worth your time.

Stay innovative,

Jackie @ AI DUNN Right

P.S. We're 340+ subscribers strong and growing. If this was useful, forward it to someone who needs better AI insights.

Got AI questions? Hit reply. I read every email.

Share this newsletter: Forward to a colleague who's tired of AI hype