Agentic Commissioning is our way of letting an AI agent work directly inside the engineering tools we already use, Siemens Process Simulate and TIA Portal, to review the PLC logic, verify that the line works and moves, and fix what is needed when it does not, before it reaches the plant floor. At RBX Programming we are building it because we believe the next step in PLC and robot programming is not writing code faster. It is proving sooner that the whole cell works.
In short
- What it is: an AI agent, equipped with MCP tools, that operates Process Simulate and TIA Portal.
- When it enters: once the PLC and the cell (trajectories included) can be tested together.
- What it does: reviews all the PLC logic, verifies that it works and that the line moves, and corrects what is needed to make it move. It looks for errors, collisions and bugs before they appear at the customer’s plant.
- How it differs from RITMO: RITMO creates and optimizes the robot trajectories. Agentic Commissioning helps us verify that everything works together.
- Shared goal: both projects aim to reduce on-site commissioning costs, shorten virtual commissioning time and improve cycle times.
What is Agentic Commissioning?
Commissioning is the stage in which a production cell is brought to a working state: the PLC logic, the robots, the sensors and the safety behaviour are tested together. Virtual Commissioning does that work in simulation, so problems are found and fixed before the equipment is on site.
The word “agentic” describes how the AI behaves. A traditional assistant suggests code or answers a question. An agent plans a sequence of steps, uses tools to carry them out, checks the result and keeps working until the task is done. In our project the agent is our AI software, and the tools it uses are the engineering applications our team works with every day.
How does Agentic Commissioning work?
Our project, led by Andrés Rodriguez, has two parts that work together.
1. MCP servers give the agent its tools
MCP (Model Context Protocol) is an open standard for connecting models to external tools and data. We use MCP servers to give our AI software everything it needs within reach to operate Process Simulate (PS) and TIA Portal (TIA). Without those tools, a model can talk about a PLC project. With them, it can work on it.
2. The agent enters at the testing stage
Agentic Commissioning starts when the PLC program and the cell, including trajectories and the rest of the simulation, have reached a state in which they can be tested together. From that point the agent reviews all of the PLC logic and verifies that it works and that the line moves. It looks for the problems that only appear when logic and motion meet: errors, collisions and bugs. If something stops the line from moving, it corrects what is needed.
In practice, the loop looks like this:
- The agent runs a test scenario on the simulated cell.
- It observes what happens on both sides, the simulation in Process Simulate and the logic in TIA Portal.
- It links any symptom to its likely cause in the PLC program or in the cell.
- It corrects what is needed in the PLC logic or in the cell so that the line moves.
- The scenario is run again to confirm the fix.
Our engineers review the changes the agent makes.
Agentic Commissioning vs RITMO: two different problems
We are often asked whether Agentic Commissioning is part of RITMO. It is not. They are two separate projects that address different moments in the life of a robotic cell, and they can complement each other.
| RITMO | Agentic Commissioning | |
|---|---|---|
| Led by | Álvaro López | Andrés Rodriguez |
| Nature | R&D algorithm (2025-2027) | Project applying AI agents through MCP |
| Key question | What are the best trajectories for the robots in a cell? | Does the PLC logic work and does the line move? |
| Main job | Creates and optimizes trajectories in cells with one or several robots | Reviews all the PLC logic, verifies the line moves and corrects what is needed |
| Project stage | Cell design and trajectory programming | Integrated testing, once PLC and cell are ready |
| Tools | NVIDIA Omniverse and Siemens Process Simulate | AI software with MCP servers for Process Simulate and TIA Portal |
| Output | Optimized trajectories | A verified PLC and cell, with the corrections applied |
| Shared goal | Reduce on-site commissioning costs, shorten virtual commissioning time and improve cycle times | |
RITMO and Agentic Commissioning side by side.
RITMO creates, Agentic Commissioning verifies.
That is the simplest way to remember it. One helps us define how the robots move. The other helps us check that, once the PLC is connected, everything behaves as it should.
Both projects pursue the same three objectives for our clients: reducing the cost of commissioning on site, reducing the time of virtual commissioning, and improving cycle times. They simply get there from different stages of the project.
RITMO is co-financed by IVACE, the Generalitat Valenciana and EU funds under the Business Value Chain Consolidation programme, and is developed with the Institute for Design and Manufacturing (IDF) of the Universitat Politècnica de València.
Benefits of Agentic Commissioning for our clients
Both RITMO and Agentic Commissioning pursue three objectives: reducing on-site commissioning costs, shortening virtual commissioning time and improving cycle times. These are the benefits we are designing Agentic Commissioning to deliver toward them. We will share measured results once we have them.
Problems found earlier, before the plant
The most expensive errors are the ones discovered during physical startup. Testing the integrated PLC and cell in simulation brings those errors forward to a stage where fixing them costs engineering time, not production time.
Broader test coverage
An agent can work through many scenarios in a systematic way, including unusual ones that are easy to skip when a team is short on time. More scenarios tested means fewer surprises later.
Less repetitive work for our engineers
Running the same checks again after every change is necessary and tiring. Delegating that loop, checks and corrections included, to an agent frees our team to spend their time on design decisions and on the problems that need judgment.
Faster feedback between PLC and simulation
Because the agent can work with both Process Simulate and TIA Portal, it can follow a symptom in the cell back to the logic that causes it, and the reverse. That shortens the path from “something is wrong” to “this is why”, and it helps shorten virtual commissioning time.
Better cycle times
Reviewing all of the PLC logic against a line that actually moves lets us spot waits and inefficiencies in the logic, which is where cycle time is often lost. Improving cycle times is one of our objectives for the project.
Corrections, not only reports
When the line does not move, the agent does not stop at flagging the problem. It corrects what is needed, which shortens the time between finding an issue and having a cell that works.
A traceable record of what was tested
An agent working through tools can leave a record of the scenarios it ran and what it found. For clients, that means clearer evidence of what has been validated before startup.
Lower commissioning costs on site
All of the above points to the same result: a smoother startup, with fewer hours spent fixing problems at the customer’s plant and fewer hidden costs appearing late in the project.
What Agentic Commissioning is not
It does not replace our engineers. The agent reviews the logic, verifies it and corrects what is needed, but our engineers design the solution and review and approve what the agent changes. It also does not replace Virtual Commissioning. It is a new way of carrying out and analysing part of it.
Why we are building it: always two steps ahead
Our tagline, “Always two steps ahead”, is a way of working: we innovate and put new technology to use so that our clients get better results. We do not adopt AI for its own sake. We apply it where it removes cost and time from real projects: fewer commissioning hours at the customer’s plant, a faster virtual commissioning stage and better cycle times.
RITMO and Agentic Commissioning are two examples. They address different stages of the same project and share one purpose: a cell that reaches production sooner and runs better.