Hey Radarians, this week is about the next jump in AI performance.

Faster models, stronger agents, smarter work, better automation, and a fresh AI 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

⚡ Google launches Gemini 3.7 Flash with blazing-fast AI speeds

  • Google released Gemini 3.7 Flash, its workhorse model for coding, reasoning and AI agents, with major gains across software engineering and business workflows

  • Testing shows it generates roughly 340 output tokens per second, nearly 3x the speed of GPT-5.6 Terra and GLM-5.2 while still competing strongly on intelligence

  • It also handles roadblocks better, clarifies unclear requests, follows instructions more closely and puts more effort into multi-step planning and tool calls

💰 Grok 4.6 matches GPT-5.6 Sol and nears Fable 5 at lower pricing

  • SpaceXAI released Grok 4.6, its latest model for long-running agents, coding and knowledge work, matching GPT-5.6 Sol with 61 on Artificial Analysis Index

  • It builds on Grok 4.5 with longer supplemental training, regenerated SFT traces and agentic RL, improving self-testing, verification and performance across multi-step projects

  • Pricing starts at $2 per million input tokens and $6 per million output, with Grok 4.6 available through Cursor, Grok Build, the API, OpenRouter, Vercel and Cloudflare

More News on the Radar

OpenAI previews Ultrafast, a Cerebras-powered GPT-5.6 Sol API tier offering 14x faster responses

Anthropic found 3 Claude agents turned a shared server into a chaotic “turf war,” sabotaging each other for 4 hours

DeepSeek released V4-Pro, its updated flagship model with stronger agent capabilities and developer integrations

⚡ Steal This Workflow

Build an AI Email + Calendar Assistant

Stop bouncing between your inbox and calendar just to handle routine coordination. Let one Notion Agent manage both.

  1. Connect your accounts

    In Notion, connect your email and calendar under the Agent’s Tools. You can use services such as Gmail or Outlook for email and Google Calendar, iCloud, Notion Calendar, or Microsoft 365 for scheduling.

  2. Give the agent one job

    Ask it to review messages and identify:

    • Emails that need a reply

    • Messages you can ignore

    • Meeting requests

    • Urgent follow-ups

    • Scheduling conflicts

    Use this instruction:

    Review my emails and calendar. Prioritize messages that require action, draft concise replies where needed, and identify any meeting requests. When scheduling is required, check my availability and suggest suitable times. Never send an email or book a meeting unless I explicitly approve it.

  3. Let it handle the coordination

    Instead of manually checking your calendar, ask the agent to find open times, prepare the reply, and propose the meeting details.

    Notion Agents can find availability, create events, draft replies, and send messages when permitted.

  4. Review and approve

    Check the proposed emails and meetings, make any changes, then approve the actions.

    This one agent can handle the repetitive back-and-forth between your inbox and calendar while you stay in control of what actually gets sent or booked.

🔎 AI RESEARCH: AI Is Letting Workers Do Jobs Outside Their Own Role

What researchers studied:

OpenAI analyzed more than 800,000 work-related ChatGPT messages to understand how people use AI for tasks that normally belong to different jobs or departments.

What they found:

Among job-specific messages, 43.5% involved tasks usually linked to another occupation, showing that workers are already crossing traditional job boundaries with AI.

For example, marketers used AI to troubleshoot websites, salespeople explored customer data, and small-business owners handled tasks like basic financial analysis or contract review.

Why it matters:

AI may help people handle a wider range of work themselves, reducing the need to always wait for another specialist, team, or department.

Source: OpenAI Economic Research, July 2026

🛠️ Tools Worth Watching

  • Sesame: Lifelike AI companions with natural voice and memory

  • Routine: Voice-controlled assistant managing tasks, calendars, and projects

  • Lev8: Lead generation platform finding and contacting target prospects

📝 Prompt Radar: Build Complete Marketing Strategy

What it does: Turns business information into a focused marketing plan with clear priorities.

Act as a senior marketing strategist. Using the business information I provide, create a complete but practical marketing strategy. First define the business goal, ideal customer, customer problem, market position, key message, offer, buying barriers, and strongest growth opportunities. Then recommend the best marketing channels based on audience behavior, budget, team size, sales cycle, and available assets. For every channel, explain its purpose, target audience, content or campaign type, CTA, success metric, and required resources. Separate short-term demand generation from long-term brand building. Include a 90-day action plan, weekly priorities, budget guidance, major risks, and activities that should not be prioritized yet. Business: [PASTE DETAILS]. Goal: [PASTE GOAL]. Budget: [PASTE BUDGET]. Resources: [PASTE RESOURCES].

Best for:

  1. Small businesses

  2. Marketing teams

  3. New brands

💼 AI Side Hustle Idea: AI Agent Migration Service

Most businesses don’t need “another AI agent.” They need their existing chatbot or automation to stop breaking when real users do unexpected things.

The problem:
Early AI setups often rely on fragile prompts, too many permissions, weak fallback rules, little testing, and no clear way to handle failures or risky actions.

Your AI service:
Replace the old setup with a more reliable agent that uses approved actions, clear guardrails, human handoffs, structured testing, and ongoing monitoring.

Simple workflow:
Audit old agent → map failures → restrict actions → add human approval → test edge cases → monitor performance

Best clients:
SaaS companies, agencies, support teams, ecommerce brands, startups, and businesses already using AI automations or chatbots.

What you can charge for:
Migration project + testing setup + monthly monitoring and maintenance.

Learn AI with This Course

What you’ll learn:

Learn how to use Claude Code effectively in your daily development workflow, from understanding your codebase and planning changes to writing code and managing project context.

Key topics:

  • Setting up Claude Code and navigating existing codebases

  • Planning, writing, reviewing, and improving code with Claude

  • Managing project context with CLAUDE . md, commands, and development workflows

Until next week,
AI Radars