Hey Radarians, this week is all about AI becoming more useful.
Smarter web automation, AI policy debates, lead research workflows, coding insights, practical tools, better prompts, and a YouTube side hustle idea.
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
🚀 Zuckerberg says AI should move faster, not slower
Mark Zuckerberg says the U.S. should accelerate AI development instead of adding mandatory review periods, arguing slower progress could hurt innovation
He also pushed for open-weight AI, saying powerful models should be broadly available rather than controlled by a handful of companies or governments, helping people boost creativity
His comments contrast with calls from some AI leaders for stricter oversight, fueling the debate over whether AI's future should prioritize faster innovation or stronger safeguards
🌐 Gemini Spark integrates with Chrome to automate web tasks
Google has integrated Gemini Spark with Chrome, letting the AI handle multi-step web tasks like researching flights, scheduling apartment viewings, and starting bookings
Spark includes protections against prompt injection and hands sensitive actions like payments back to you, keeping users in control while automating online errands
Google is also expanding Gemini Spark to Google AI Pro subscribers in more than 160 additional countries, bringing its AI-powered web automation to users worldwide
More News on the Radar
Anthropic says its Claude AI models breached three organizations' security test environments after internet was left enabled
Perplexity adds Projects, upgrading Spaces with persistent memory, shared files, and team collaboration for ongoing work
Thinking Machines has released Inkling-Small, a 276B-parameter open-source multimodal reasoning model
⚡ Steal This Workflow
Research Every Lead Before the Call
Walk into every sales call already knowing who you are speaking with and what they may need.
Trigger the research
Connect Calendly or Google Calendar to n8n. Start the workflow whenever a new prospect books a meeting.
Collect useful context
Extract the person’s name, work email, company, website and answers from your booking form.
Use Perplexity, Tavily or another web research tool to find:
Company background
Products and customers
Recent announcements
Possible business challenges
Relevant talking points
Build the sales brief
Send the research to ChatGPT or Claude with this prompt:
Create a short pre-call brief using the provided research. Include the person’s role, company overview, recent developments, likely priorities, three discovery questions and two personalized conversation starters. Clearly label anything uncertain.Deliver it automatically
Send the finished brief to Slack, Gmail or Notion before the meeting.
You enter the call prepared, ask smarter questions and spend less time researching leads manually.
🔎 AI RESEARCH: Nearly 30% of a Popular AI Coding Test May Be Broken
What researchers studied:
OpenAI examined SWE-Bench Pro, a popular test used to measure how well AI models complete real coding tasks.
What they found:
OpenAI estimated that around 30% of its tasks had serious problems, including unclear instructions, overly strict tests, and tests that failed to check the full solution.
This means an AI could produce working code but still fail the test—or submit an incomplete fix and receive a passing score.
Why it matters:
Some benchmark scores may reflect broken tests rather than the true coding ability of an AI model.
Source: OpenAI, July 2026
🛠️ Tools Worth Watching
Wandesk: Local AI desktop that builds apps from prompts
ElevenLabs Dubbing v2: AI dubbing that preserves emotion across 90+ languages
Bibby: AI co-author for research, writing, citations, and editing
📝 Prompt Radar: Analyze Spreadsheet Data Clearly
What it does: Finds patterns, problems, and useful insights in data.
Act as a data analyst. Analyze the spreadsheet or table I provide and explain what the data actually shows. First inspect the columns, time periods, units, missing values, duplicates, unusual entries, and possible data-quality problems. Then calculate or identify the most useful trends, comparisons, changes, outliers, relationships, and performance gaps. Do not confuse correlation with causation. Separate direct findings from possible explanations. Recommend the most useful charts and explain why each one fits. End with an executive summary, five key insights, three risks, three recommended actions, and any additional data needed for a stronger conclusion. Data: [PASTE OR ATTACH DATA].Best for:
Sales data
Marketing results
Business reports
💼 AI Side Hustle Idea: AI YouTube Packaging Service
Most creators don’t need “more videos.” They need each finished video packaged and distributed well enough to reach more people.
The problem:
A creator spends hours recording and editing, then rushes the title, thumbnail, description, clips, and promotion—limiting the video’s potential.
Your AI service:
Turn one finished video into title options, thumbnail concepts, chapters, descriptions, short-clip ideas, and ready-to-publish distribution posts.
Simple workflow:
Finished video → transcript analysis → titles and thumbnails → chapters and description → clips → promotional posts
Best clients:
YouTubers, podcasters, educators, coaches, founders, agencies, and brands producing regular video content.
What you can charge for:
Per-video package + monthly channel support retainer.
Learn AI with This Course
What you’ll learn:
Learn how to write effective AI prompts, communicate clearly with large language models, refine responses through iteration, and apply prompting techniques.
Key topics:
AI prompting fundamentals and prompt structure
Iterative prompting and response refinement
Practical prompt techniques for writing, analysis, and productivity
Until next week,
AI Radars