AI is moving beyond screens—and closer to your operations
You may already use software for accounting, customer messages, stock records, or staff scheduling. But many business tasks still depend on people physically seeing, moving, checking, sorting, and responding to things. A warehouse worker locates an item, a technician inspects equipment, or a café team member checks whether supplies have arrived.
A new wave of artificial intelligence is being built to handle that physical world. The reported fundraising by General Intuition shows how investors are backing AI systems designed to understand movement, space, timing, and actions—not just text. The company is reportedly discussing funding at a $6 billion pre-money valuation, after raising $320 million at a $2.3 billion valuation just weeks earlier. Source: TechCrunch
You do not need to buy a robot or build an AI laboratory to pay attention. The practical lesson is simpler: business software will increasingly help your team understand and act on real-world work. That makes this a useful time to improve your processes, records, and operating discipline.
TL;DR
Physical AI combines intelligent software with cameras, sensors, machines, or robots that can interpret and act in real spaces. General Intuition is reportedly developing a foundation model for agents that can move through space and time, with a focus on robotic applications. Source: TechCrunch
For your SME, the immediate opportunity is not replacing staff. It is identifying repetitive physical tasks, capturing reliable operational data, and connecting automation to work your team already performs.
What This Means
Most AI tools you encounter today work with information on a screen. You provide a document, message, image, or question, and the system produces an answer. Physical AI aims to go further by helping machines understand what is happening in a physical environment and decide what action should happen next.
For example, a system could use camera feeds and sensors to identify a box, estimate its position, recognise that an aisle is blocked, and instruct a machine or worker on the next step. In a service environment, it might observe equipment behaviour and flag an unusual condition before a breakdown. The important distinction is that the AI is connected to movement and time, not only written instructions.
General Intuition reportedly began with data from Medal, a video game clip-sharing platform founded by its chief executive Pim de Witte. The company used hundreds of millions of hours of gameplay and “action labels”, which record what buttons players pressed and when, as an initial dataset. Source: TechCrunch
That approach matters because intelligent action requires more than recognising an object. A useful system needs to understand sequences: what happened first, what changed, what action worked, and what should happen next. The company reportedly plans to improve its general model with robotic embodiments, while investing in computing infrastructure and additional talent. Source: TechCrunch
The lesson for your business is not “buy a robot now.” It is “make your real-world work clear enough for technology to understand and improve.”
How This Applies to Malaysian SMEs
1. Warehousing and distribution. If you operate a small warehouse, online shop, spare-parts business, or wholesale operation, much of your daily work involves locating goods, checking quantities, preparing orders, and identifying damaged stock. Physical AI may eventually help cameras and mobile machines recognise products, navigate storage areas, and confirm whether an order has been prepared correctly. You can prepare by using consistent product codes, labelled shelves, barcode scanning, and a simple record of stock movements.
Start by listing the physical steps in one common order. Record where mistakes occur: wrong item selection, incomplete packing, misplaced stock, or delays in receiving goods. This gives you a practical baseline for deciding whether ordinary workflow automation, better scanning, or future camera-based tools would help. The aim is not to automate everything. It is to remove avoidable checking and searching so your people can handle exceptions.
2. Retail, food service, and hospitality. A shop, restaurant, bakery, or small hotel has many activities that happen repeatedly: checking shelves, monitoring queues, restocking supplies, inspecting cleanliness, and preparing standard items. Sensors or cameras could eventually identify low stock, unusual movement, unsafe conditions, or a workstation that needs attention. In the near term, you can gain similar benefits by using digital checklists, scheduled alerts, and photo-based confirmation for important tasks.
For example, instead of relying on memory for closing procedures, create a digital sequence covering equipment, storage, cleaning, and security. Require a staff member to confirm each step and record exceptions. This creates action data: what was done, when it was done, and what needed follow-up. That information can later support more advanced systems, while already improving consistency today.
3. Maintenance, construction, and field services. If your team installs air-conditioning units, services machinery, handles electrical work, maintains facilities, or manages building projects, employees often inspect physical conditions and make decisions while moving between locations. AI-assisted vision could eventually help identify visible defects, compare conditions over time, and guide technicians through standard procedures.
You can prepare without specialised equipment by standardising inspection forms. Ask technicians to capture the same photos from similar angles, record symptoms using defined categories, and note the action taken. Over time, these records can help you identify recurring faults, training needs, and common delays. They also make it easier to review work quality when the owner or manager is not present at every site.
4. Small-scale manufacturing and packaging. Physical AI may be relevant to businesses handling assembly, packaging, printing, or quality inspection. A vision system could eventually identify missing components, incorrect labels, or damaged items while products move through a work area. Before considering such technology, measure how quality checks currently happen and separate simple rules from genuinely difficult judgement.
If a worker checks the same label placement hundreds of times, that is a candidate for a clearer standard or assisted inspection. If every product is different and requires experience, automation may need to support the worker rather than replace the decision. Your first step should be to document acceptable and unacceptable examples, including photographs where appropriate.
Useful preparation for your business
| Business area | Physical task | Preparation step | Useful measure |
|---|---|---|---|
| Warehouse | Pick and pack orders | Use consistent item codes and shelf labels | Picking errors per 100 orders |
| Retail or café | Restock and closing checks | Create a timed digital checklist | Missed checks per week |
| Field service | Inspect and repair equipment | Standardise photos and fault categories | Repeat visits per job |
| Manufacturing | Quality inspection | Capture clear examples of pass and fail items | Defects found after dispatch |
These measures are operational indicators, not forecasts. They help you understand whether a process is stable enough for automation and where human attention is most valuable.
Practical Takeaways
- Map one physical workflow. Choose a repeated process such as receiving stock, preparing an order, or inspecting a service job.
- Record actions, not just outcomes. Note what was checked, moved, approved, rejected, or repaired.
- Standardise names and locations. Consistent item codes, shelf labels, fault categories, and task descriptions make future automation easier.
- Measure exceptions. Count wrong picks, repeat visits, missed checks, and damaged items rather than relying on general impressions.
- Keep people responsible for judgement. Use automation to assist with routine work while your team handles unusual, sensitive, or safety-related decisions.
- Review privacy and safety. Cameras, location records, and worker activity data require clear internal rules and careful access control.
- Test small improvements first. A digital checklist or barcode process may solve the immediate problem without requiring advanced robotics.
The Bigger Picture
The General Intuition story points to a wider direction in technology: AI developers are trying to create systems that can understand actions across different environments, rather than perform only one narrow task. The reported involvement of investors including Valor Equity Partners, Point72 Ventures, Seven Seven Six, Khosla Ventures, and General Catalyst reflects strong interest in this area. Source: TechCrunch
For Malaysian SMEs, adoption will probably be gradual. You may first see better cameras, inspection software, warehouse tools, delivery systems, and equipment-monitoring services rather than humanoid robots walking into small businesses. The providers that succeed will need to work with local conditions, including mixed-language teams, different layouts, uneven connectivity, and processes that are less standardised than those of large corporations.
Your advantage is that you can make decisions close to the work. You know which tasks consume staff time, where errors affect customers, and which processes depend too heavily on one experienced employee. Begin there. Build reliable procedures, collect useful records, and introduce automation only where it supports a clear business need.
Physical AI may take years to become practical for many smaller companies, but preparation starts with ordinary management: clear workflows, consistent data, measurable quality, and staff who understand why a new tool is being introduced. When the technology becomes accessible, businesses with disciplined operations will be in a stronger position to use it safely and effectively.
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 →