Request a quote
Benešl Automation / AI for TIA Portal
Artificial intelligence for Siemens TIA Portal · S7-1500 · S7-1200 · WinCC Unified

AI for TIA Portal: an assistant that sees your project

Want to use ChatGPT, Copilot, Claude or Gemini for Siemens PLC programming? TIA MCP connects your AI assistant directly to the project open in TIA Portal. The AI writes a block in SCL, LAD or FBD, compiles it, fixes the errors, reviews the program and writes the documentation – in the project, not in a chat window.

What AI in TIA Portal looks like

A plain-language request, the result right in the project: the AI creates a function block in SCL, fixes the warnings after compiling and adds the call in LAD. A real recording from TIA Portal V21, only shortened.

FB in SCL, fixing warnings after compiling and the call in LAD (38 s, no sound)

Why copying code from ChatGPT is not enough

A chat on its own does not know your project. It writes generic code that you then adapt to your tags by hand, paste into TIA Portal and fix after compiling. With TIA MCP the same AI works directly in the project.

AI chat on its ownAI with TIA MCP
Sees your project – blocks, tags, UDTs, hardwareno, only what you paste into the chatyes, reads directly from TIA Portal
Follows your names and conventionsguessesyes, reads them from the project
Compiles the code and fixes compile errorsno, you find the errors yourselfyes, in one loop
Writes the result into the projectno, you copy it by handyes, with a backup before every change
LAD, FBD, WinCC Unified HMI, PROFINETtext onlyyes, as TIA MCP tools

What you can do with AI in TIA Portal

You give instructions in plain language, the AI picks the TIA MCP tools itself and shows the result in the project. You check and approve.

Writing SCL, LAD and FBD blocks

FBs, FCs, DBs and UDTs from a plain-language description or a specification. The AI compiles the block and fixes the errors itself.

Program review

Code review of a block or the whole PLC, unused objects, multiple writes, safety program checks and naming conventions.

Documentation and manuals

Software handover documentation, block descriptions, I/O lists, block help and an operating manual in PDF with HMI screens.

WinCC Unified

Screens, faceplates, tags, dynamizations and scripts. Migration of Comfort panels to Unified and automated HMI testing in simulation.

Migration and conversion

STL to SCL, projects from Beckhoff TwinCAT, B&R, CODESYS or Schneider into TIA Portal, and library upgrades from older versions.

Why the machine is down

A service technician asks in plain language and the AI walks through the program, fills in live values from the CPU and points to the device and terminal in the schematic.

Hardware and PROFINET

Devices, modules, I/O addresses, IP plan and PROFINET names, network topology and an I/O checklist for commissioning.

Project translations

Alarm texts, comments and screens into other languages with a glossary and a check that nothing was left untranslated.

All 452 tools by area

Example requests

This is how you really work with AI in TIA Portal. No commands or syntax – just write what you need.

Write an FB to control a pump with dry-run protection and a fault alarm.

Review the PLC_1 program and list what needs fixing before commissioning.

Convert the STL blocks to SCL and check that they do the same thing.

Create the software handover documentation for the customer as a PDF.

Why is conveyor 3 stopped? Go through the program and tell me what to check.

Translate the alarm texts into English and German.

What MCP is – in simple terms

MCP (Model Context Protocol) is a standard that lets AI assistants such as ChatGPT, Claude or Copilot use tools in other programs. TIA MCP is a set of such tools for TIA Portal: it gives the AI “hands” in the project – it can read blocks, tags and hardware, create and change objects and run the compiler.

Install TIA MCP

The Windows installer finds your AI clients and registers TIA MCP in them. Nothing to configure by hand.

Open the project in TIA Portal

TIA Portal V17 to V21 on the same computer. TIA MCP connects to it through the official Openness interface.

Write what you need

In ChatGPT, Claude, Copilot or the TIA MCP web interface. Writes to the project are backed up and you can approve them beforehand.

Which AI you can use

The one you already have. TIA MCP works with Claude, ChatGPT, GitHub Copilot in VS Code, Microsoft Copilot Studio, Gemini, Cursor and other MCP clients. For working offline also with a local model on your computer (Ollama, LM Studio) – the project then never leaves your network.

ChatGPTClaudeGitHub CopilotCopilot StudioGeminiCursorOllamaLM Studio

AI in the project, but under control

A backup before changes

Before a block, tag or screen is changed, a backup is saved and the change can be undone with one command.

Read-only mode

You switch off writing to the project with one toggle, and only a person can switch it back on, never the AI. You can approve changes one by one.

Runs on your computer

The TIA MCP server runs on your computer and the licence is checked offline. Downloading to a real PLC is disabled by default.

In more detail: what AI can do in TIA Portal

The most common reasons people look for AI in TIA Portal, described in detail with example prompts.

SCL generator

FBs, FCs, DBs and UDTs from a plain-language description, following your conventions, compiled and checked against S7-1500 instructions.

How the SCL generator works

Migrating S7-300/400 to S7-1500

What TIA Portal does and where AI helps: STL to SCL, old system functions, absolute addresses, comparison and tests.

The migration procedure to S7-1500

AI for WinCC Unified

Screens, faceplates, dynamization, tags and alarms, migration from Comfort panels and HMI testing.

AI and WinCC Unified screens

Testing in PLCSIM

Unit tests of blocks, test suites, PLCSIM scenarios and a virtual FAT with a report.

How to test a PLC program with AI

Project documentation

Operating manual, handover documentation, IO lists, block documentation and change reports.

Documentation with AI

Git and project versions

The project as text in Git, comparison, history, backups and restoring blocks.

Git for TIA Portal

Training for your team

A workshop on site or online: safe setup, writing blocks, review, documentation, migration, service and local AI.

Training content and format

FAQ about AI in TIA Portal

Can AI program Siemens PLCs?

Yes, if it can see the project. With TIA MCP the AI writes SCL, LAD and FBD blocks for S7-1500 and S7-1200, compiles them in TIA Portal and fixes compile errors. The programmer gives the instructions, checks and approves – the AI is an assistant, not a replacement for experience.

Does ChatGPT or Copilot work with TIA Portal?

Not on their own – they cannot see into TIA Portal. Through TIA MCP they can: ChatGPT, Claude, GitHub Copilot, Microsoft Copilot Studio, Gemini and Cursor then work directly with the open project. TIA MCP also adds a Copilot right into TIA Portal (right-click on a block).

Is TIA MCP a Siemens product?

No. TIA MCP is independent software from Benešl Automation. It talks to TIA Portal through the official TIA Portal Openness interface, which is part of the TIA Portal installation, and contains no Siemens components.

What do I need?

A Windows computer with TIA Portal V17 to V21 with Openness and the AI client you already use. The installer sets up the rest.

Do I need internet? Will my project go to the cloud?

TIA MCP runs locally and checks the licence offline. A cloud AI assistant receives only the data it requests for the task. If you do not want to send anything out, use a local model (Ollama, LM Studio) – then you need no internet at all.

How much does it cost?

A PRO licence costs €295 per year per computer, excluding VAT. You can try TIA MCP free for the first 7 days.

What if the AI makes a mistake?

The code is always compiled in TIA Portal, so the AI sees compile errors and fixes them. A backup is made before every change and you can restore it with one command, and nothing is downloaded to a real PLC without your explicit confirmation. Always check the program logic before commissioning.