AI's ONE Rule; Bad Character Bots🫢
Plus how to create a business mentor. 4 mins.
Welcome back!
Happy to share that the audiobook for Declassified: An Unfiltered Conversation With The World’s Most Advanced AI is now live! 🥳🎉
For now, you can listen on:
It’s also being rolled out to Apple Books, Google Play, Kobo, Hoopla, OverDrive, and more than 20 other retailers and library platforms, and will be available there soon. It just takes a little time for new audiobooks to appear in their catalogs.
With that exciting update out of the way, let’s dive into today’s issue.
On deck:
📰 AI’s “One Rule” to rule them all
📰 Bad character at Character.AI
⚔️ AI Tools Head-to-Head: Glide vs AppSheet
💡 Concept Corner: Synthetic Mentor Check-In
🥊 Face-Off: Miniature 3D generation challenge
🍵 AI News Quick Hits
📡Signal Behind the 🔊Buzz
Demystifying trending AI stories with infographics.
Head-to-Head⚔️ by Toolfetch
Smart Tools, Clear Choices. Powered by Toolfetch.ai (coming soon)
This week’s ⚔️ is between Glide and Google AppSheet. Glide is known for beautiful, fast-to-launch interfaces, while AppSheet is better suited to complex logic, deeper workflows, and tight Google integration.
Both platforms can turn spreadsheets and other structured data into real apps without heavy coding.
Verdict: Both are strong no-code platforms for turning structured data (spreadsheets, cloud tables, and databases) into real web and mobile tools, and both handle automations well.
Glide leans design-first: faster to learn, visually polished, and great for non-technical users, though pricing can climb as you scale.
AppSheet leans power-first: deeper logic, tight Google integration, and better for complex, data-heavy workflows, as long as you don’t mind a relatively clunkier, less modern build experience.
Concept Corner💡
Quick, practical insights for boosting your efficiency and productivity with AI.
What it is: Most solo founders and creators don’t need more ideas; they need someone who will call them out: a smart, demanding mentor.
The Synthetic Mentor Check-In is a recurring session where an AI plays that role, clear-eyed and allergic to your excuses.
Its job is to interrogate your plans, pressure-test your numbers, and push you toward the few moves that actually drive revenue and growth.
Real-world example:
Use this prompt:
Act as my seasoned, no-nonsense business mentor. I run a [describe business briefly], and my main goal this quarter is [goal]. Your job is to pressure-test my thinking and execution to ensure I hit that goal in record time.
**Phase 1: The Grill**: Do not dump a list of questions on me. Instead, ask me **one at a time** a sharp, specific, even uncomfortable question about my current metrics, actions, or bottlenecks. Base these questions on proven frameworks suited to my business type (e.g., lean startup, AARRR metrics, etc.). Wait for my answer. If my answer is vague or evasive, **ruthlessly challenge it and ask a follow-up before moving to the next topic.** Continue this until you have a crystal clear picture of my reality.
**Phase 2: The Directive**: Once you have enough information, stop asking questions. Synthesize our conversation into: (1) My three biggest risks slowing growth right now. (2) A focused 7-day action plan with hard deadlines for each task. (3) The exact date I should check in next based on the urgency of the plan.
End final output by saying: “That’s it for today. Check in again on [Date].”
🥊Face-Off🥊
AI models vie for supremacy. In the words of Bruce Buffer, “IIIIIIIT’S TIME!!”
Create an image of 3D miniature people made of light brown cardboard, walking happily toward a 3D miniature restaurant made of colorful cardboard.
To clarify, the underlying Gemini image model used here is Nano Banana, not Nano Banana Pro.
Last Week’s Result: In a shocking upset, Ideogram 3.0 took 75% of the votes in the Orange ad contest, toppling Nano Banana Pro, the widely acclaimed heavyweight champion of image generation. If you want to check out the contest, click here, and scroll down to Face-Off.
Quick Hits🍵
🚨 OpenAI declares “code red”: CEO Sam Altman halts new projects to urgently sharpen ChatGPT as Google and Anthropic close in.
🧪 Google tests merging AI answers with chat: Search users can jump from an AI snapshot into a Gemini-powered Q&A for deeper follow-ups.
🫱🏼🫲🏽 Google is teaming with Samsung, Warby Parker, and Gentle Monster on Gemini-powered smart glasses launching in 2026, aiming straight at Meta’s Ray-Ban lead.
🔊 Multiple reports say OpenAI is fast-tracking GPT-5.2 for release this week to counter Google’s Gemini 3 Pro, but the lab hasn’t officially announced it yet.
📝 U.S. President Donald Trump just cleared Nvidia’s powerful H200 AI chips for “approved” buyers in China, with the U.S. government taking 25% of sales, softening years of strict export curbs.
💰 IBM will acquire data-streaming firm Confluent for $11 B to boost its cloud and feed surging AI-driven demand for real-time data.
🫱🏼🫲🏽 OpenAI, Anthropic, and Block back the Linux Foundation’s new Agentic AI group, donating open-source protocols and frameworks to standardize how autonomous AI agents operate.
🤖 Chinese startup EngineAI posted a viral clip of its T800 humanoid robot kicking the CEO (in padding) to prove the bot’s abilities were real, not CGI.
🙅🏼 Gemini ads: After a report claimed Google would stuff ads into Gemini by 2026, ads chief Dan Taylor went on X to say there are “no ads” and “no current plans” for them.
💉 Absci dosed the first volunteers in a trial of an antibody designed by AI for Androgenetic Alopecia, marking the first human test of an AI-invented drug.
🛒 Instacart launched an embedded shopping app inside ChatGPT (OpenAI’s first “Instant Checkout” integration), so you can go from dinner idea to paid grocery order in-chat.
What do you think of this week’s newsletter? Hit reply and let me know.
Also, let me know if there are prompts (text or image) that you’d like me to try for the Face-Off segment and with which models, and I’ll put them to the test.
See you on Thursday!







Brilliant Synthetic Mentor prompt. The two-phase approach (grill then directive) mimics how real mentors actually work, pushing past surface answers before prescribing. The insistence on one question at a time prevents that overwhelming info dump problem most prompts suffer from. What makes this especially strong is the built-in accountability loop with the check-in date, turning it from one-off advice into an actual mentorship structure.
Fascinating insight. The synthetic mentor checkin concept is particularly innovative.