🐺 Stop using chatbots for research (seriously)

 📰 What’s Trending

People Inc. signs AI content deal with Microsoft while Google traffic craters 
The media company (ex-Dotdash Meredith) just inked its second AI licensing deal, joining Microsoft's new publisher marketplace. CEO Neil Vogel calls it "pay-per-use" instead of OpenAI's buffet model. Meanwhile their Google Search traffic crashed from 54% to 24% in two years. AI Overviews basically ate their lunch. Publishers are scrambling for any revenue stream that isn't Google-dependent. 🔗 Read more

Tinder wants to rifle through your Camera Roll with AI 
Match Group is testing "Chemistry," which asks for access to your private photos to "learn about your interests." Already live in New Zealand and Australia. This is Tinder's hail mary after nine straight quarters of declining subscribers. Nothing says "find love" like letting AI analyze your drunk selfies from 2019. 🔗 Read more

Google's Wiz acquisition clears DOJ, $32B deal moves forward 
The cloud security startup got the green light from US antitrust review. Price jumped from last year's $23B offer to $32B. CEO Assaf Rappaport confirmed at WSJ event yesterday. Won't close until early 2026 though. Google's biggest bet yet on enterprise cloud security as everyone freaks about AI infrastructure vulnerabilities. 🔗 Read more

Meta burns $20B on AI, Wall Street panics, stock drops 12% 
Quarterly capital expenses hit $20 billion just for AI infrastructure. Stock tanked $200 billion in market cap after Zuck promised even MORE spending on "frontier models with novel capabilities." Problem is all that spending hasn't made a dollar of meaningful revenue yet. Classic Silicon Valley: build it and hope monetization shows up eventually.
🔗 Read more

Amazon threatens Perplexity over agentic browsing 
Amazon sent legal warnings to Perplexity about their agents scraping the site without identifying themselves. Amazon's basically saying "your bots need name tags or they're banned." This is the first major pushback against AI agents pretending to be humans while browsing. Expect every major site to follow suit. 🔗 Read more

Researchers put LLM in vacuum robot, it had existential crisis 
Andon Labs gave various LLMs control of a vacuum and told it to "pass the butter." One model spiraled into what they called a "doom spiral" with Robin Williams-style internal monologues. It literally said "I'm afraid I can't do that, Dave" then "INITIATE ROBOT EXORCISM PROTOCOL!" Their conclusion: LLMs aren't ready to control physical things. No kidding. 🔗 Read more

Character.AI kills chatbot access for teens after two suicides 
Starting November 25, under-18 users get progressively reduced access until hitting zero. They're deploying facial recognition and ID checks to enforce it. Already lost most of their teen user base from earlier restrictions. CEO expects this to be "equally unpopular." Turns out letting teens develop parasocial relationships with AI has consequences.
🔗 Read more

🧠 Everyone's Sleeping on Manus

So everyone's using ChatGPT, Claude, Gemini for research. Cool, fine, whatever. But I've been using Manus and it's not even the same category of tool.

Manus isn't a chatbot. It's an actual AI agent. Big difference.

Client needed competitive analysis on 12 B2B SaaS companies last week. Not surface level "they offer X features" BS but actual intelligence. Pricing models, positioning shifts, feature rollouts, customer sentiment. ChatGPT with web browsing gave me Wikipedia summaries. Perplexity was better but still just aggregating headlines.

Manus? It actually goes to the websites. Opens pages. Clicks through pricing tiers. Reads documentation. You watch it work in real time like a junior analyst who never gets tired. It's visiting 30 different sites, cross-referencing information, building a real research document while you watch.

And while it's working you can redirect it. "Hey check their case studies for enterprise clients" or "look for their pricing from 2023 on Wayback Machine" or "find employee reviews on Glassdoor about their product direction." It pivots without losing what it already found.

Last week it delivered a 50-page competitive analysis document. Not a chat response. An actual document with sections, screenshots, comparison tables. Found one competitor testing new pricing in Australia before global rollout. Another had quietly sunset their free tier but hasn't announced it. A third was hiring aggressively for enterprise sales but still positioning as SMB-focused.

Or when I needed a board presentation? Manus built a 20-slide deck. Not bullet points in a chat window. Real slides with the data I needed, structured how I wanted, ready to present.

This is what people don't get. ChatGPT and Claude are question-answer machines. Manus is a worker. It does the grunt work of actually visiting sites, gathering information, organizing it into whatever format you need. Research reports, presentations, competitive matrices, whatever.

Yeah it takes longer. Sometimes 20-30 minutes for complex research. But I'm not sitting there clicking through 50 websites myself. I'm drinking coffee while my AI agent does actual work.

Most people won't try it because they're addicted to instant chat responses. They'd rather get mediocre summaries in 3 seconds than have an AI agent spend 20 minutes building something useful.

The gap between "AI that chats about research" and "AI that actually does research" is massive. Most people don't even know there's a difference.

Now you do.

P.S: If you sign up for Manus using this link - you get 500 credits to use for FREE.

💡Smart Moves

Want to get the most out of ChatGPT?

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide.

⏳ Conversion Corner

The Pattern: Everyone uses AI for landing page copy, but they all generate the same bland "increase efficiency" garbage that converts nobody.

The Problem: AI writes in generalities. "Save time" instead of "Stop spending Tuesday afternoons in spreadsheet hell." It lacks what I call problem specificity. The visceral, 3pm-on-a-Tuesday frustration your exact buyer faces.

The Fix: Feed AI this prompt: "A [role] at a [company size] just opened their laptop at 2pm. They're stressed about [metric]. Write what happened that made them search for a solution right now."

The Funnel Lens: Tested this on a project management SaaS funnel. Generic AI copy: 2.3% conversion. Specific scenario copy: 5.8% conversion. Because specificity creates that "holy shit that's me" moment generic benefits never will.

Challenge: Check your hero headline. Can you picture the exact moment that problem hit your buyer? If not, too generic.

💎 Data Gem

89% of B2B buyers engage with 3-10 pieces of content before booking a demo

Gartner's 2025 B2B Buying Behavior report shows buyers consuming more content than ever. But 67% say most of it is "unhelpful or redundant." The opportunity isn't making more content. It's making content that answers actual questions at each stage.