The multiplatform AI agent
specialized for PLC logic.
Connects directly to your IDE: CODESYS-based PLCs and Beckhoff TwinCAT.
1FUNCTION_BLOCK FB_Conveyor2VAR3 speed : REAL;4END_VAR56IF speed > 0.0 THEN7 motor.Run();8END_IF
Full support for the PLC platforms you already run
A bridge between the chat and your IDE.
A small bridge runs on your machine next to CODESYS or TwinCAT. The chat sends it instructions; the bridge writes them into your open project and streams the code, the compile results and the errors back. The loop closes by itself — the agent fixes what does not compile.
- Download it from the app and run it next to your IDE
- It attaches to the project already open — nothing is imported or exported
- Your project stays on your machine; only the conversation leaves it
Writes Structured Text like a controls engineer.
Fine-tuned on IEC 61131-3 — function blocks, scan-cycle idioms and vendor-specific dialects. It reads your existing blocks, GVLs and data types before it writes a line.
- Reads your function blocks, globals, data types and installed libraries first
- Takes your vendor manuals and coding standards as input
- Output matches your project's conventions, not a generic template
Reads your spec. Implements it.
Upload a PDF specification and the agent turns it into function blocks, logic and state machines — wired into the project you already have open.
- Reads the PDF and works from the section you point it at
- Creates the blocks, the state machine and the wiring between them
- Compiles what it wrote before handing it back
Compiles, reads the errors, fixes them.
After every change the agent compiles your project, reads the vendor's own compiler output and resolves the errors on its own. You review the final result, not the intermediate steps.
- Vendor diagnostics, verbatim — the compiler is the oracle, not the model
- Fixes are proposed as a diff, like any other change
- Re-compiles to prove the fix rather than claiming it
Catches edge cases before they ship.
Reviews your code for quality issues, missed edge cases and risky patterns — like a senior engineer looking over your shoulder before it goes on the machine.
- Missing interlocks, unbounded values and dead branches
- Checked against your own standards, not generic lint rules
- Every suggestion comes as a diff you can take or leave
You approve every change.
See exactly what the assistant proposes before it touches the project. Accept, reject, or edit. The build result comes back from the vendor's own compiler, unmodified.
- A diff per object, shown before anything is written
- Vendor diagnostics reported verbatim — no summarised 'looks fine'
- Upload your library docs and coding standards so generated code matches
Latest articles
Insights, updates and stories from our team.
AI Tools for PLC Programmers: What's Available in 2026
A practical overview of AI tools that PLC programmers can use today -- from general-purpose chatbots like ChatGPT and Claude to specialized CODESYS copilots. What works, what doesn't, and how to choose.
10 PLC Programming Best Practices for Clean Code
Best practices for writing professional PLC code in Structured Text. Covers naming conventions, state machines, error handling, modularity, and testing strategies used by experienced automation engineers.
AI Chatbot for CODESYS: Write PLC Code by Talking
What if you could describe your PLC logic in plain English and get working Structured Text code back? AI chatbots for CODESYS are making this possible. Here's how it works and what to expect.
Bring your PLC project into the modern workflow.
Start free. CODESYS and Beckhoff TwinCAT.