A warehouse robot project looks like a hardware purchase and turns out to be an integration project. The robots take their work from your warehouse system and report back every pick, short and empty bin, all day, and your ERP has to act on what they report. If it can only receive "done", the robots' information ends up in a spreadsheet, and the engineers who own the ERP inherit the gap.
Robotic automation in the supply chain mostly means robots that move goods around warehouses. The International Federation of Robotics counted 102,900 professional service robots sold for transport and logistics in 2024, up 14% on the year before and the largest category of professional service robot, most of them mobile robots that carry and handle goods. Amazon deployed its millionth robot in 2025, across more than 300 facilities.
For a mid-size distributor or manufacturer, the open question is whether your warehouse software and your data are ready to tell a robot what to do, and to believe what it reports back.
Updated by the TechFabric editors in October 2026.
Two different things called robotic automation
Writing on the subject often mixes up two technologies that share a word.
Robotic process automation (RPA) is software. A bot drives an application's screens the way a person would, clicking and typing, to move data between systems that have no integration. In Microsoft's world this is a desktop flow in Power Automate, which automates repetitive desktop processes through an application's UI, with data entry into an ERP as Microsoft's own example. It's useful, cheap to start, and has nothing to do with moving pallets.
Physical robots are machines. In a warehouse that means autonomous mobile robots (AMRs) that bring shelves or totes to a picker or follow pickers through aisles, automated storage and retrieval systems (AS/RS) that store goods in dense grids and deliver them to a port, conveyors and sorters, and, increasingly, robot arms that pick individual items.
Both need the same thing from you, which is reliable data about what's where. They differ in what happens when that data is wrong. A software bot fails and somebody re-keys an order. A robot drives to an empty location.
The scale of it
Some context on how fast this is growing, from IFR's World Robotics 2025 figures:
- 542,000 industrial robots were installed worldwide in 2024, more than double the number ten years earlier.
- The operational stock reached 4.66 million.
- The United States installed 34,200, down 9% on 2023.
Treat any robotics statistic more than two years old as out of date; the market moves faster than the articles about it.
What a robot needs from your warehouse system
Robots don't decide what to do. A warehouse system tells them, through a layer the industry calls warehouse execution or warehouse control software, usually supplied by the robot vendor. The conversation between the two runs in both directions, all day:
- Down to the robots: which work exists, which order lines, from which location, to which station, in what priority.
- Up from the robots: what was actually picked, what was short, which location was empty, which licence plate (pallet or tote ID) arrived where.
The second direction is where projects get hard. When a robot reports a short pick, something in your ERP has to adjust the inventory, decide whether to pick from another location, and tell customer service. If your system can only receive "done," the robot's information ends up in a spreadsheet.
What Dynamics 365 provides
Dynamics 365 Supply Chain Management has a documented interface for exactly this, the material handling equipment interface (MHAX). It works through two queues. Outbound events tell the equipment when warehouse work is created, started, completed or cancelled, and when picks and puts are done. Inbound events let the equipment confirm work, report short picks, override a location and receive licence plates. The equipment side polls for events through a SOAP or REST service, so any vendor that integrates with it uses the same contract. It became generally available in April 2021.
Supply Chain Management also offers warehouse management only mode, where a legal entity runs only the warehouse and takes shipment orders from another ERP or order management system. That's relevant if your ERP is something else but your warehouse is about to become automated and needs a WMS that can talk to the equipment.
Business Central is different. It has three levels of warehouse handling, the most advanced being directed put-away and pick, with mandatory bins, zones, bin ranking and capacity. We haven't found a documented equivalent of MHAX for it, so robot integration with Business Central runs through its APIs and a custom integration that maps the robot vendor's messages onto warehouse documents. That's a reasonable build, and it's worth scoping before the robots are on order rather than after.
What to fix before the robots arrive
Most of the work in a warehouse automation project is data, and most of it can start today:
Bin-level inventory accuracy. Robots route to locations. If your system says a bin holds 40 units and it holds 32, a person shrugs and picks 32. A robot raises an exception. Run cycle counts on your fastest-moving locations for a month and measure how often the system was right.
Item dimensions and weights. Totes, shelves and grids have size and weight limits. Every item that goes into automated storage needs accurate length, width, height and weight in the item master, and in an older item master those fields are often blank or years out of date.
Licence plates. If pallets and totes don't have IDs that your system tracks, the robots can't report against them. In Business Central that usually means moving to directed put-away and pick, with bins, before the automation project, and that's a project in itself.
Exception handling. Write down what should happen for a short pick, a damaged item, a robot that's down, and a rush order that has to jump the queue. Each of those needs a system behaviour, not just a person who knows what to do.
Choosing where to start
A building doesn't need to be automated all at once, and the order matters more than the brand of robot.
Goods-to-person AMRs, which bring storage to a picker, suit operations with many small order lines and a large range of items, where walking is most of the picking time. Person-following AMRs, which carry totes alongside pickers, change less of the building and suit a trial in one zone. AS/RS suits dense storage of small, fast-moving items where floor space is expensive. Conveyors and sorters suit high, steady volumes of cartons.
Whichever fits, a pilot in one zone with a defined measure, order lines per labour hour and pick accuracy, compared against the same zone before, tells you more than any vendor's case study.
The robots are the visible part. The ERP, the warehouse system and the data behind them decide whether they work. TechFabric implements Dynamics 365 Business Central for warehouse teams and builds the integrations between the ERP and the systems around it.