The Robot News That Almost Went Over Your Head
You run a business in Malaysia—maybe a food supplier, a spare parts distributor, a cleaning service. You wake up to rising costs, staff who don’t show up, and orders that pile up. The last thing on your mind is what Google’s AI division did with a humanoid robot last week. And yet, this particular piece of tech news isn’t about a lab experiment. It’s about a machine that can bend down, pick up a watering can, and tie a trash bag without being told exactly how to move each joint.
If you’ve ever watched a worker in your warehouse stop halfway through picking an order because they had to stretch or kneel, you already understand the problem these robots are trying to solve. Previous robots could wave their arms around but couldn’t safely coordinate the rest of their body. That’s now changing. And while you won’t buy one tomorrow, the capability roadmap matters for your business planning.
TL;DR
Google DeepMind’s Gemini Robotics 2 model now controls humanoid robots from “feet to fingertips,” enabling walking, crouching, stretching, and better hand dexterity. It also lets multiple robots work together and run offline. For Malaysian SMEs, this signals that physical labour automation is becoming more practical—so start standardising your processes now.
What This Means (Without the Tech Jargon)
Think of a robot like an office worker. The old version could only use their arms—like someone sitting at a desk all day. The new version has full body control: it can stand up, walk over to a shelf, crouch, pick something up, and then place it exactly where you want. The latest announcement says the model now supports “whole-body motions” ranging from feet to fingertips.
The upgrade also touches hands. Gemini Robotics 2 controls “more complex, five-fingered hands,” meaning it can do tasks like sealing a Ziploc bag, tying a trash bag, or unscrewing a lightbulb. These aren’t flashy demos—they’re the kind of fine motor skills that every business has somewhere in its operations.
There’s also an upgrade to the “embodied reasoning” model, which helps robots understand their surroundings and plan multi-step tasks. It “now understands when tasks begin and end”—so it knows the difference between picking up a box and finishing the whole delivery route. Additionally, the on-device model can run without an internet connection, which matters if your factory or storage area has spotty Wi-Fi.
How This Applies to Malaysian SMEs
Let’s get grounded. You’re not Google. You’re not a robotics company. But you are someone with repetitive physical tasks that eat into your margins. In Malaysia, SMEs make up 97.4% of businesses, and many rely on manual labour for warehousing, food prep, packaging, and retail stocking. The question isn’t whether robots are cool. It’s whether they’ll become cheap enough and adaptable enough for a 20-person operation.
The “whole-body” feature matters because it removes a big barrier to robotic adoption in real Malaysian environments. Most SME spaces are not designed for robots. Narrow aisles, uneven flooring, shelves at different heights, and mixed inventory. Earlier robots with only upper-body control would struggle to reach down or step over clutter. Gemini Robotics 2’s ability to walk, crouch, and stretch means it can handle your messy storage room, not just a pristine lab. One video from Google shows a robot bending over to pick up a watering can and finding specific items on a shelf. That’s the difference between robots that need a custom-built facility and robots that can work in a converted shoplot in Puchong.
Consider your food and beverage supply chain. Many Malaysian SMEs receive bulk ingredients, portion them, and pack them for restaurants. Sealing plastic bags and tying trash bags are exactly the kind of five-fingered hand tasks the new model can handle. Even a simple task like unscrewing a lightbulb could be part of a maintenance checklist. If a robot can maintain your prep area, you free up a human worker to focus on quality checks and customer relationships—things that can’t be automated.
Then there’s teamwork. The announcement highlights how different types of robots can cooperate, with one robot instructing another to put tools in a bin while cleaning a garage. Small businesses rarely need one superhuman robot; they need four or five cheap machines doing different jobs that complement each other. This update moves toward that reality. The same AI model can run on a wheeled robot in your front office and a humanoid robot in your back store—sharing the same “brain” and coordinating tasks.
“The next wave of automation isn’t about keyboards—it’s about hands, feet, and balance. SMEs that prepare their processes now will be the ones who actually benefit when the price drops.”
Practical Takeaways for You
- Audit your physical tasks. List every job that requires bending, reaching, or fine hand movement. These are the prime targets for whole-body robots within the next 3–5 years.
- Standardise your workplace layout. Robots do best when things are consistent. Start labelling shelves clearly and keeping floors clear today. That costs little, but it makes automation easier later.
- Watch local integrators. You won’t buy directly from Google. Malaysian automation companies will package this technology into affordable solutions. Follow them, not just tech news.
- Don’t rush to buy. The announcement itself notes robots “have more to advance in movement speed.” Wait for second or third-generation models, and rent or lease first if possible.
- Train your current staff on basic AI concepts. When robots arrive, you’ll need people who can supervise them and troubleshoot simple errors—not programmers, just comfortable users.
What the Capability Roadmap Looks Like
| Feature | Previous Model | Gemini Robotics 2 | What It Means for SMEs |
|---|---|---|---|
| Body control | Upper body only | Full body, feet to fingertips | Can work in real spaces with floors and obstacles |
| Hand dexterity | Simple grippers | Five-fingered hands | Can tie bags, seal packaging, unscrew items |
| Task awareness | Step-by-step commands | Understands task start and end | Guides better inventory management |
| Multi-robot coordination | No | Yes, different robot types work together | Parallel processing for different parts of operations |
| Internet connection | Required | On-device, works offline | Works in remote warehouses and factories |
Source: The Verge’s coverage of Google DeepMind’s announcement
The Bigger Picture
Here’s what you need to understand: this is not about a cool robot demo. It’s about the cost of labour vs. the cost of machines crossing a line. For decades, automation required either massive capital investment or rigid setups that only made sense for huge factories. Whole-body AI models change that calculus because they can handle variability—the messiness of real work.
For Malaysian SMEs, the long-term trend is clear. Your competitors in Singapore and Thailand will adopt these robots faster in some cases, and you’ll feel pressure on both labour costs and delivery speed. But you also have an advantage: smaller teams mean easier adoption. You can test one robot in one branch, prove it works, and scale it without corporate bureaucracy.
The window to prepare is now. No need to buy anything. Just start observing your own processes and asking: “Which parts of this job require common sense about the human body?” Those are the parts that will soon become automated. When a machine can tie a trash bag, it can do a lot of other things too.
Ready to Streamline Your Operations?
Your business should run itself. AutoRunBiz deploys AI agents to automate your daily operations — WhatsApp orders, invoicing, customer follow-ups, and accounting. Book a free 15-min ops audit to see where automation fits your business →