😸 That $20K robot? It's a guy with a VR headset

PLUS: All the top open (AI) model drops from October!

Welcome, humans.

This is an admittedly somewhat misleading chart showing the S&P 500 soaring while job openings tank right after ChatGPT's launch that went viral on Reddit this week:

Reddit being Reddit, the top comment immediately called it misleading (no y-axis scales = chart crime 101).

Here's what's actually going on: job openings started dropping in 2022, right when the Fed hiked interest rates oh just about seven times to fight inflation, plus big tech over-hired during COVID and had to course correct.

So while the divergence looks dramatic, it's probably 95% economic fundamentals and 5% AI impact; though to be fair, Fed Chair Jerome Powell did say AI is “definitely impacting jobs.”

The real story = AI hype is propping up the market (thanks NVIDIA!), while the broader economy is... let's say, less excited.

Correlation doesn't equal causation (yet), but it sure makes for a spicy chart!

Here’s what happened in AI today:

NEO released a humanoid robot that’s actually a mechanical skin suit.
NVIDIA secured 260k+ Blackwell chip orders for South Korea.
OpenAI launched Aardvark for automated code vulnerability scanning and patching.
Meta and xAI started using off-balance-sheet debt for AI infrastructure.

Advertise in The Neuron here

Meet NEO, the humanoid robot that's not quite ready for your chores.

Last week, 1X Technologies launched NEO, a sleek humanoid robot promising to automate household chores (for a cool $20K). Imagine a robot that tidies your house, does your laundry, and empties the dishwasher while you’re at work…according to this demo video, it can do it all!

But hold on. According to tech reviewer MKBHD, the product as seen in the launch video is actually “not real”; at least, not in the way you think.

In a new video, he breaks down the massive gap between what NEO is promised to do and what it can actually do. The stunning demos of NEO folding clothes and organizing kitchens? Almost all of it is teleoperated, meaning a human in another room is controlling the robot with a VR headset.

Here's the reality check:

The Big Reveal: MKBHD points to a new WSJ report where 100% of the live demos shown to journalist Joanna Stern were remote-controlled by a human.
Limited Autonomy: In its nearly 10-minute launch video, 1X only labels two simple tasks as truly autonomous: opening a door and taking a single cup from someone's hand. Asking for a friend: how big of a party would you have to throw to have $20K worth of red solo cups lying around to make Neo worth it?
Selling the Dream: This strategy is becoming a trend in AI hardware. MKBHD compares it to the Humane Pin and Rabbit r1, who both promised a revolutionary future while selling a work-in-progress. Spoiler: They both flopped.

Now, the CEO calls this all a “bumpy adventure.” In his interview with the WSJ, 1X CEO Bernt Børnich was surprisingly frank about the robot's limitations:

He confirmed that early adopters are signing up for a “social contract.”
The deal = you let remote operators help NEO, providing the data needed to make the AI smarter.
Børnich promised privacy controls like blurring faces and user-approved
no-go zones,” but admitted, “If we don't have your data, we can't make the product better.”

Ooooh okay, so basically they’re not just tele-operating the robo for the demo, they plan to tele-operate the robot INSIDE YOUR HOUSE. This is freaky for a variety of reasons, but the most futuristic of which could lead to some Star Wars level spy dramas, where your personal droids report back on your private convos to the Empire.

Even in his own home, the CEO says NEO currently saves him about “half an hour per day” by doing some tidying and vacuuming. Unloading the dishwasher? That only works “on a good day,” and getting the door autonomously fails about 10% of the time.

The CEO's honesty confirms MKBHD’s core point: the AI isn't there yet. This “sell the dream, build it later” model asks customers to take huge financial and privacy leaps of faith for some dude to do your chores from home. What are we even doing here, y’all?

Anyway, for early adopters with deep pockets and a borderline problematic risk tolerance, you can pre-order NEO now for $20K upfront or a $499/month subscription (it’ll begin shipping sometime in 2026). But be warned: for now, NEO is a high-tech human-operated demo, NOT a fully autonomous butler.

Oh, and if you’re looking for robots you can buy now, check these out: Booster K1 is a tiny robot that costs $6K (video), while Bumi is an even smaller robot that costs only ~$1.4K (video). The robot bodies are definitely here now. The brains? Still TBD.

FROM OUR PARTNERS

AI Without Trust Is Just Faster Mistakes

AI is revolutionizing work—but most companies are realizing a harsh truth: their AI is only as smart as the knowledge behind it.

That’s where Guru comes in. Trusted by thousands of enterprises, Guru is the AI Source of Truth that connects all your data, then delivers cited, permission-aware answers you can trust. No hallucinations. No conflicting docs.

Just verified, compliant knowledge everywhere you work—Slack, Teams, browser, and all your AI chat tools and agents. Because in the age of AI, truth isn’t optional—it’s protection.

Watch a demo

Prompt Tip of the Day

Harvard Professor and author Arthur Brooks went on CNBC last week to promote his current book, the “Happiness Files”, when the topic of AI and its impact on happiness came up, and what he said REALLY made sense to us.

See, Brooks just wrote a book about Happiness, and according to his research, the happiest people focus on four things that AI fundamentally can't replace: faith, family, friends, and work that feels like a calling.

However, he has ANOTHER new book coming out next March about how we've fundamentally changed our brains with technology. The issue = because we’re always using tech, we're overusing our left brain hemisphere (things and tasks) while neglecting our right hemisphere (mystery and meaning).

How does AI fit into this mix? Here’s what Arthur says

The AI Happiness Framework:

✅ DO use AI for: Automating left brain “tasks” to free up time for right brain meaning, i.e your kids, spouse, and real world relationships.
❌ DON'T use AI as: Your therapist, friend, or confidant. Brooks warns: “Your brain knows it's being used incorrectly, and you'll get sad, anxious, and lonely.”

To us, this means you should think of AI as your left-brain assistant: great for drafting emails and content, generating code snippets you’d otherwise be scouring Stack Overflow for, analyzing data for trends, or organizing and summarizing large chunks of information (don’t forget to add “with 100% fidelity to the original” 😉).

But the moment you finish those tasks? Close the laptop. Use that reclaimed time for what ACTUALLY makes us humans happy: messy, meaningful, face-to-face connection.

Now, if you do use AI for emotional support, try this prompt approach: Instead of asking AI to “help me feel better about X” or “give me advice on my relationship / this situation,” think about what all you need to do today that’s stressing you out, make a list of tasks AI can assist you with instead, then have it knock out those lesser stressors so you can chat with your actual friends / family on those more important “meaning” issues.

Using AI to buy back time for the stuff that matters. That’s the REAL productivity hack.

Treats to Try

Today’s treats will all be AI coding related; tomorrow, we’ll share a broader mix of content for you!

OpenAI launched Aardvark, which scans your code repositories to find security vulnerabilities before attackers do, validates they're real by testing them, and generates one-click patches you can review and apply.
Amp Free lets you autonomously build features and fix bugs across your codebase with agentic coding, switching between free (ad-supported) and paid modes per task (The Rate Limited podcast covers the true code of “FREE” AI coding; worth listening to if this interests you).
1. Rate Limited also covered Cursor 2.0, and Ray Fernando ALSO did a live stream with Luke Alvociro from Factory to chat about Droid, which handles full coding tasks (refactors, migrations, incident response) across your IDE, Slack, and terminal without switching tools up to about 7M tokens (which is wild y’all). Luke’s take on where AI coding goes from here at 1:06:10 is particularly worth a watch.
OpenCode is an open-source terminal coding agent you can use with something like GLM-4.6 (via API here), giving you a 200K context window and stronger code generation while keeping full control over your models and data.
Linear has a new Slack Agent that creates issues from your team discussions; just mention @Linear in any Slack thread and it'll automatically pull the context and turn the conversation into a properly formatted Linear issue.
Check out this AWESOME list of open-source AI projects released this month.

Around the Horn

NVIDIA lined up 260k+ Blackwells for South Korea — biggest hard-numbers signal of global AI build-out momentum.
Meta and xAI now use off-balance-sheet debt to finance the AI boom.
MediaTek is targeting multi-billion accelerator revenue by 2027 — credible challenger story beyond Nvidia.
Sean Goedecke says AI’s em‑dash overuse likely stems from digitizing content from the 1800s and early 1900s, which used ~30% more em-dashes overall.
Apparently YouTuber PewDiePie maxxed out his home computer setup with 8 GPUs (a $20K setup), and now runs protein folding simulations overnight to help science research (and apparently you can do this too if you have a home computer via Folding @ Home…check it out here).
- Also, he actually has a pretty good breakdown on local models, where the biggest model he was able to run was Qwen 235B (quantized version) via vLLM (which this link explains well).
- Oh, and btw: if you want to delete your ChatGPT chats and instruct OpenAI NOT to keep your data, we learned from this video you go here (privacy.openai.com/policies) and select “make a Privacy Request.” 🌈 The More You Know 💫
- Last thing: his trusted robot council is WILD and hilarious (basically an agentic mixture of experts set-up); at one point he figured out he could run 64 different AI models at one time (Qwen 2.5 3B Instruct) with RAG and search.

FROM OUR PARTNERS

CTV ads made easy: Black Friday edition

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting — plus creative upscaling tools that transform existing assets into CTV-ready video ads. Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

Sunday Special

This compilation of “AI chiropractor videos” (a.k.a chiropractor-ing gone out of control) is the exact definition of why they invented the phrase: “Kids, PLEASE don’t try this at home”