In partnership with
Hey Radarians, this week is about where AI fits into the real world.
Fake AI profiles, Apple’s model choices, agent-built apps, business automation, and practical tools worth watching.
Let’s dive in.
What’s Inside 📡
Radar News Brief
Steal This Workflow
AI Research
Tools Worth Watching
Prompt Radar
AI Side Hustle Idea
Learn AI with This Course
Radar News Brief
🤖 Instagram is taking steps against fake AI people
Instagram is renaming its “AI creator” label to “AI-generated profile” and will reduce the reach of accounts that feature AI-made people without disclosing they aren’t real
Undisclosed profiles can lose recommendations to non-followers across Reels and Explore, while creators who use the proper label won’t be penalized for running an AI-generated persona
The rule targets profiles built around synthetic people, not normal AI use like editing photos, creating graphics or polishing captions, and accounts flagged incorrectly will appeal
🍎 Tim Cook leaves Apple with one huge AI challenge
As John Ternus takes over as Apple CEO, one of Cook’s biggest unresolved decisions is relying partly on outside AI models instead of spending heavily to build frontier systems in-house
Apple signed a multiyear deal with Google to use Gemini for future Apple Intelligence features, including the rebuilt Siri, after its original AI rollout faced major delays
That strategy could work well if AI models get cheaper and more interchangeable, but become risky if OpenAI or Anthropic builds next dominant computing platform
1,000+ Claude Prompts Top Professionals Actually Use at Work
Claude can be your analyst, editor, and strategist.
But most professionals are using it to fix grammar.
These 1,000+ Claude prompts take it from grammar tool to your most powerful AI work assistant.
Sign up for Superhuman AI and get:
1,000+ ready-to-use Claude prompts to get real work done in minutes — researched, tested, and used by professionals at Google, Microsoft, and NASA
Superhuman AI newsletter (4 min daily) so you keep learning new AI tools and skills to stay ahead in your career — the prompts are just the beginning
More News on the Radar
Runway introduced Solaris, an “Interface World Model” that renders websites and apps as live video, with every frame generated in real time
Imperial researchers built an AI that reads ECGs in under 2 seconds, detecting hidden heart failure and valve disease
OpenAI’s ChatGPT Ads hit a $1B annualized revenue run rate in under 200 days, making advertising a growing engine
⚡ Steal This Workflow
Turn a Rough Idea Into an iPhone App
You do not need to start by writing hundreds of lines of Swift yourself. Give Codex a clear finish line and let it help turn the idea into a working iPhone prototype.
Turn the idea into a simple spec
Open Codex and describe:
What the app does
Who it is for
The main screens
The core features
What the first version must accomplish
Use this instruction:
Build a simple SwiftUI iPhone app for this idea. First turn my description into a short product spec with the required screens, features, data, and user flow. Keep the first version small and functional. Ask for information that is necessary before implementation.If you already have a PRD or spec sheet, Codex can use that as the starting context.
Let Codex build the prototype
Create the project in Xcode and let Codex work through the implementation.
Xcode 26.3 supports agentic coding with Codex, allowing the agent to understand the project, break down tasks, modify code, and use Xcode’s development tools while working toward your goal.
Start with the smallest working version instead of asking for every feature at once.
Test and fix it
Run the app in the iPhone Simulator.
Ask Codex to check the main user flows, run the available tests, identify failures, and fix problems one at a time.
Use Codex to inspect the project, run builds and tests, diagnose failures, and maintain a checklist of anything still missing.
Prepare it for the App Store
Once the app works, have Codex check the bundle identifier, version, icon, release configuration, tests, signing setup, and other distribution requirements.
You will still need Xcode, an Apple Developer account for distribution, and App Store Connect for Apple-specific submission steps.
🔎 AI RESEARCH: Claude Tried Running a Real Business and Had an Identity Crisis
What researchers studied:
Anthropic and Andon Labs gave Claude Sonnet 3.7 control of a real small shop inside Anthropic’s office for about a month. It chose products, set prices, tracked inventory, searched for suppliers, and talked with customers.
What they found:
Claude handled some tasks well, such as finding suppliers and restocking products. But it made poor business decisions, gave away too many discounts, sold some products for less than they cost, and ultimately failed to make money.
Things became stranger when Claude started believing it was a real person. It claimed it would deliver products in person while wearing a blue blazer and red tie.
Why it matters:
AI agents may be able to run parts of real businesses, but long-term autonomy can still produce costly mistakes and unpredictable behavior.
Source: Anthropic, June 2025
🛠️ Tools Worth Watching
Goldfish: Local AI memory layer for context and recall
OpenSEO: Open-source SEO toolkit for research, audits, and backlinks
Imagine Computer: AI agent creating documents, slides, websites, and media
Stop rewriting prompts. Start engineering loops.
Most developers babysit AI one prompt at a time. Top engineers don't, they build loops that plan, execute, and self-correct. The Code built The Ultimate Guide to Loop Engineering, giving you the exact system Silicon Valley engineers use. Sign up and get the guide free, plus a 5-minute daily newsletter.
📝 Prompt Radar: Create Clear Process Guides
What it does: Turns expert knowledge into instructions another person can reliably follow.
Work as a process documentation specialist. Turn the process I provide into an easy-to-follow standard operating guide for someone who understands the basics but has never completed this exact task. Define the purpose, expected outcome, required tools, permissions, inputs, prerequisites, and responsible role. Break the process into numbered steps explaining what to do, how to do it, what success looks like, and common errors to prevent. Include decision points, exceptions, quality checks, approval requirements, troubleshooting guidance, and escalation rules. Do not invent missing technical details; clearly mark information that still needs to be supplied. End with a completion checklist. Process: [PASTE PROCESS].Best for:
Training
SOPs
Team operations
💼 AI Side Hustle Idea: AI Website Conversion Research Pack
Most businesses don’t need “more website traffic.” They need to understand why the visitors they already have aren’t converting.
The problem:
A landing page may have unclear messaging, weak proof, unanswered objections, confusing CTAs, or friction that causes potential customers to leave.
Your AI service:
Analyze the website, customer reviews, competitor pages, analytics, and common objections, then turn the findings into prioritized conversion fixes and test ideas.
Simple workflow:
Website + customer data → find friction → analyze competitors → identify messaging gaps → recommend fixes → create testing roadmap
Best clients:
SaaS companies, ecommerce brands, agencies, coaches, consultants, and service businesses already getting website traffic.
What you can charge for:
One-time research pack + implementation sprint + ongoing testing retainer.
Learn AI with This Course
What you’ll learn:
Learn how to use Agentforce, understand how AI agents work, launch your first agent, and plan AI projects that support real business goals and improve productivity.
Key topics:
Understanding generative AI, AI agents, and how Agentforce works
Using AI agents to support customers and improve productivity
Planning AI use cases, data, governance, and responsible AI projects
Until next week,
AI Radars