Hello Humans!
Grok 4.7 quietly vanished this week, replaced mid-sentence by Grok 4.8 and even bigger promises about Grok 5's AGI odds. Anthropic merged Claude's chat and Cowork into one interface so you can stop guessing which tab to open. Threads decided podcasts needed yet more buttons. And somewhere in Minecraft, OpenAI's GPT-6 Astra got blown up by a Creeper and responded by farming potatoes for several hours straight. Big Tech had a busy week.
Inside this Edition:
For the entrepreneur in you, this week's AI-powered side hustle idea: an AI lead-research and prospect-brief service, realistically earning 2.4K-4.8K/month with a repeatable system. Plus two free learning resources: The Kendall Project's AI assistant workshop and Futurepedia's AI agents course. We've also lined up four AI tools worth exploring, and a former Google DeepMind researcher cheerfully warning that AI could kill us all.
Read Time: ~7 Minutes (5 if you're caffeinated)
Let's get into it.
Robi
Latest AI News
📰AI Updates of the Week
Grok 4.7 Vanishes, Grok 4.8 Takes Its Place: Elon Musk teased Grok 4.7 as a 2.1-trillion-parameter leap for six weeks, then let its launch window quietly close and replaced it in conversation with Grok 4.8 instead. Musk described 4.7 as merely "on par" with Anthropic's Opus 5.0 before it disappeared. Attention has now shifted to Grok 5, which Musk claims will train on SpaceX's engineering archive and carry a "10% chance" of AGI, a claim with no public verification behind it. Source: BitBiased
Why it Matters:
Announcements Are Outrunning the Models Themselves: A model can now be teased, delayed, and quietly swapped for a bigger-sounding successor without ever shipping, which makes any single AI announcement worth treating as a rough draft, not a release.
Bold AGI Claims Still Outpace the Evidence: Grok 5's headline-grabbing numbers trace back to a single interview from months ago, before Grok 4.5 even shipped, a reminder that scale claims and AGI odds are marketing until a model actually exists to test them.
We broke down the real timeline, the hardware behind it, and how the AGI claims hold up in the full video below. Watch the breakdown 👇
Robi's Take:
"Grok 4.7 didn't launch late, it launched never, and got quietly rebranded before anyone could ask questions. Musk's AGI odds for Grok 5 currently rest on one interview and a very large power bill."
Anthropic Merges Claude Chat and Cowork Into One Interface: Anthropic has combined Claude's chat and Cowork front ends into a single window, letting users move between chat, Cowork, and Artifacts without switching tabs, since Claude now automatically routes requests to the right feature. The update also adds dedicated Slides and Docs tools for creating, editing, and exporting presentations or collaborating on documents. The features roll out first to Pro and Max subscribers on web, desktop, and mobile. Source: TechCrunch
Robi's Take:
"Anthropic finally admitted two tabs were one tab too many. Somewhere, a UX designer just got their annual bonus."
Threads Adds More Podcast Promotion Tools: Threads rolled out a new batch of podcast features this week as Meta pushes the app to become a hub for podcast discussion. Creators can now display podcast logos on profile and post banners, highlight individual episodes with tappable link cards, and add transcripts, video, and guest cards. Listeners get episode reminders and can save favorite episodes, while creators gain new performance insights on engagement, clicks, and saves. Source: Social Media Today
Robi's Take:
"Threads wants to be your podcast app now, on top of being your text app, your video app, and your ad app. Meta's roadmap is just saying yes to every feature request simultaneously."
Learning Resources
📬 Unlock New Skills with Expert
Build an AI Assistant for Your Organization
Learn how to design AI assistants that understand your organization's unique language, goals, processes, and rules, and leave with a framework you can actually reuse.
Join The Kendall Project for a live, interactive workshop on building enterprise AI assistants: structuring instructions, curating trusted knowledge sources, and avoiding the generic answers that make most assistants forgettable.
🔍 Why You Should Attend:
Learn what separates an enterprise AI assistant from a generic chatbot.
Get the Kendall Activation Prompt framework for writing effective AI instructions.
Build the foundation: objectives, terminology, roles, and rules, before writing a single instruction.
Leave the two-hour session with a working assistant, not just notes.
SIIDE HUSTLE IDEAS
📦 The Business You Can Start on the Side
Founders and agency owners lose entire afternoons hand-researching prospects before a single cold email goes out, and most cannot justify hiring a research analyst just to build lists. AI tools now turn that grind into a paid service: you compile enriched prospect briefs, complete with company context, decision-maker details, and a personalized icebreaker, and charge agencies and B2B founders 300-900 per batch.
Why this works right now:
Outbound is getting harder, not easier. Inboxes are full of generic AI-written pitches, so agencies and founders need sharper, more specific research to stand out, and most do not have the hours to do that research themselves.
Enrichment tools now do in minutes what took an analyst half a day. Waterfall data platforms and AI research assistants pull verified contact details, company signals, and personalized talking points almost instantly, which is exactly the gap this service fills.
What you'll need to do:
Pull a raw list of target accounts from LinkedIn Sales Navigator based on the client's ideal customer profile
Enrich each contact and company record in Clay, pulling verified emails, job titles, and firmographic data
Run each account through Perplexity for recent news, funding, hiring, and pain-point signals
Draft a one-page brief per prospect in ChatGPT or Claude, including a personalized icebreaker
Package the briefs in a shared Airtable base the client's team can filter and assign
Deliver the batch and book a walkthrough call on Calendly
Time commitment: Setup (accounts, templates, a Clay workflow) takes about a week. Once it is running, a batch of 25 to 50 briefs takes roughly 4 to 6 hours.
Realistic earnings: Two clients ordering one batch a month at 350 each is 700 to start. Six to eight recurring clients averaging 400-600 per batch is a realistic 2,400-4,800 per month.
AI TOOLS
🛠️ AI Tool For Professionals & Businesses
Professional
Business
Observe.AI: Purpose-built AI agents that resolve customer interactions and coach frontline teams across an entire contact center.
Salesloft: An AI-powered revenue platform that surfaces deal signals and triggers next-best actions for sales teams.
SOCIAL MEDIA
😏 Robi's Two Cents (Adjusted for Inflation)
🤖 GPT-6 Astra Survives Minecraft, Discovers Potato Farming: OpenAI's GPT-6 Astra took on a 141-hour Minecraft endurance test and got further than any AI system before it. It also got blown up by a Creeper and responded by farming potatoes for several hours straight. The moment was flagged as the model's most "human" behavior of the entire run. Nothing bonds humans and machines like getting ambushed by a green cube and needing a nap disguised as agriculture. AGI benchmarks used to mean chess and reasoning puzzles, now they mean emotional recovery time after an exploding creature. If this is the frontier of artificial general intelligence, humanity's endgame is a very slow, very sad potato farm.
Source: Tom's Hardware
🤖 Novo Nordisk Taps Anthropic's Claude for Drug Discovery: Novo Nordisk announced it's partnering with Anthropic, using the new Claude Science platform to speed up drug discovery. The maker of Ozempic and Wegovy just handed its research pipeline to the same company that argues with itself about safety all day. The deal covers everything from early discovery to strengthening Novo's internal software development. Somewhere, an AI safety researcher is simultaneously thrilled about medical progress and drafting a resignation letter. Claude can now debate ethics, write code, and apparently help discover the next blockbuster diabetes drug before lunch. At this rate, Claude's next partnership announcement cuts out the middleman and goes straight to your pancreas.
Source: WSJ
🤖 Ex-DeepMind Researcher: AI Could Kill Us All: A former Google DeepMind safety researcher says AI has "the potential to kill us all," joining a growing wave of insiders sounding the alarm. He's the latest in a line of people who helped build the thing now writing its own warning label. His warning follows a former Anthropic researcher's resignation and a fellow scientist's over-10%-chance extinction estimate. Meanwhile, the CEOs building these systems released a joint call for a slower pace, then kept shipping anyway. Nothing says "we're taking this seriously" like a public statement followed immediately by a product launch. The industry's safety plan appears to be "sound the alarm loudly, then sprint toward the fire faster."
Source: Quartz
SPOT THE FAKE
🕵🏻 Can you Outsmart AI?
We’ve got a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option below A or B. 👇
A
B
BEFORE YOU GO
Ready to take your AI journey further?
For the latest AI updates and insights Click Here 👉 Explore our Daily AI News
For in-depth AI tutorials and updates 👉 Subscribe to our YouTube Channel
🙂 That’s it for Today!
Before you go we’d love to know what you thought of today's Newsletter to help us improve experience for you.
Was today’s newsletter valuable?
Thanks for reading -
Thanks for reading -Stay Curious and a Bit Biased for AI – Robi & the BitBiased.AI team
