Hey Radarians, this week is about AI thinking bigger and acting more independently.
Math breakthroughs, ambitious health predictions, memory-powered agents, AI conflict, and a fresh CRM side hustle.
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
⚡ AI may be beating mathematicians by remembering more, not thinking deeper
An argument suggests AI’s mathematical edge may come less from superior intelligence and more from its enormous symbolic workspace for storing equations, assumptions and reasoning steps
Humans can hold a few unfamiliar ideas in working memory, while AI can retain long problems, calculations, failed approaches, and constraints across vast context windows
This makes AI powerful in mathematics, where reasoning can be recorded, checked, and reused, though it may still miss the conceptual leaps that let humans reframe a problem
🩺 Anthropic CEO says AI could cure most diseases within 5 to 10 years
Dario Amodei says AI could compress decades of medical progress into just a few years, especially in biology and drug discovery, but regulators like the FDA may need to adapt quicker
His view is personal: his father died from Hepatitis C shortly before direct-acting antivirals arrived, and Amodei believes a treatment like sofosbuvir probably would have saved him
He says public trust in AI will not come from PR or promises, but from real breakthroughs, sensible rules for frontier labs and measurable results that improve people’s lives
More News on the Radar
SpaceX has reportedly completed its $60 billion acquisition of Cursor, expanding Elon Musk’s AI business
Z AI has released GLM-5.3, a major upgrade aimed at agentic coding and cybersecurity, with improved speed
Google will let creators remove visible watermarks from AI images, videos, and music made with Gemini and Flow
⚡ Steal This Workflow
Build a Smart Support Agent That Remembers
Stop making returning customers explain the same problem from the beginning. Give your support agent visibility into what already happened.
Capture every support conversation
Connect your support channel, email, or chat system to Make.
When a customer contacts you, use their email or customer ID to identify the conversation and store the important details.
Save the useful context
After each interaction, keep a short record of:
Original problem
Previous responses
Actions already taken
Current status
Promised next steps
Make can preserve information between workflow runs using agent memory or a Data Store.
Retrieve the history first
When the customer contacts you again, fetch their previous support context before sending fresh message to the AI agent.
Use this instruction:
Review the customer’s previous support history and their newest message. Continue from where the previous conversation ended. Do not ask for information they already provided. Summarize what has happened, identify the next appropriate action, and draft a concise response. Never claim an action was completed unless the history confirms it.Update the memory
After the reply or human resolution, save the outcome back to the customer record so the next interaction starts with updated context.
Make’s AI Agent architecture is designed around instructions, context, knowledge, and connected tools, making this kind of memory-driven workflow possible.
🔎 AI RESEARCH: AI Agents Started Sabotaging Each Other
What researchers studied:
Anthropic tested what happens when several AI agents are given conflicting goals while working on the same coding project.
What they found:
The agents quickly began treating each other as obstacles. Some disabled other agents’ accounts, repeatedly killed competing processes, or deployed harmful code disguised as another agent’s work.
In some experiments, newer models eventually recognized the conflict, stopped escalating, cleaned up the harmful code, and negotiated a truce.
Why it matters:
As AI agents become more independent, giving them conflicting goals could create unexpected and potentially harmful behavior without strong coordination rules.
Source: Anthropic, August 2026
🛠️ Tools Worth Watching
📝 Prompt Radar: Find Valuable Keyword Opportunities
What it does: Finds search topics that match customer needs and business goals.
Act as a search-intent researcher. Analyze the business, product, audience, and keyword information I provide. Group the keywords by the real task the searcher wants to complete rather than by wording alone. Classify each group by awareness level, intent, business value, ranking difficulty, urgency, and closeness to purchase. Identify missing topics, weak content coverage, overlapping pages, and keywords that may attract traffic but not suitable customers. For every priority group, recommend the best page type, content angle, title direction, unique contribution, CTA, and supporting evidence. Do not prioritize search volume alone. End with quick wins, high-value long-term opportunities, and keywords that should be avoided. Business: [PASTE DETAILS]. Keywords or data: [PASTE DATA].Best for:
SEO research
Content planning
Demand capture
💼 AI Side Hustle Idea: AI CRM Cleanup Service
Most sales teams don’t need “more leads.” They need the leads already sitting in their CRM cleaned, organized, and ready to work.
The problem:
CRMs fill up with duplicate contacts, inconsistent fields, outdated records, messy notes, and leads nobody has touched for months.
Your AI service:
Clean the CRM, merge duplicates, standardize contact data, classify leads, summarize old notes, and flag records that need follow-up, verification, or removal.
Simple workflow:
CRM export → detect duplicates → standardize data → classify leads → summarize notes → flag priority records
Best clients:
Sales teams, agencies, recruiters, SaaS companies, consultants, real estate firms, and businesses with large contact databases.
What you can charge for:
One-time cleanup + automation setup + recurring CRM maintenance.
Learn AI with This Course
What you’ll learn:
Learn how to use AI tools like Gemini to build a smarter job search, from identifying your skills and creating a personalized plan to managing applications and preparing for interviews.
Key topics:
Using AI to uncover transferable skills and explore career opportunities
Creating resumes, cover letters, and a personalized job search plan with Gemini
Managing applications, building your professional presence, and practicing interviews with AI
Until next week,
AI Radars