# PLCs.ai

Extended reference for AI assistants (intent-to-URL map, full capability and pricing detail): https://www.plcs.ai/llms-full.txt

> PLCs.ai is an AI platform purpose-built for PLC (Programmable Logic Controller) engineers. It ingests complete Allen-Bradley (Studio 5000), Siemens (TIA Portal), and CODESYS V3 projects, reasons across every PLC on a production line, and confirms root causes against live tag values streamed from the running controller. Upload a project and understand, troubleshoot, document, and generate code for it in plain English.

## What we do

PLCs.ai is the first AI platform purpose-built for PLC engineers. It turns complex PLC code into plain English — making it possible for anyone in a manufacturing or industrial operation to understand, troubleshoot, update, and document PLC programs, regardless of who wrote them or how long ago.

## Who we serve

- Controls engineers working with legacy or modern PLC code
- Plant managers who need visibility into production line logic
- Operations directors managing multi-site manufacturing
- Maintenance technicians troubleshooting downtime in real time

## Core capabilities

- **Explain**: Upload any PLC project file and get a plain-language explanation of what it does, how it works, and how components relate to each other
- **Troubleshoot**: Ask questions about faults, logic, and behaviour in natural language — no coding knowledge required
- **Generate**: Produce safe, context-aware PLC code modifications from natural language prompts, with review before any change is saved
- **Document**: Automatically generate documentation across lines and plants
- **Version**: Snapshot code versions, changes, and insights — nothing gets lost when engineers leave

## Supported PLC vendors

- Rockwell Automation / Allen-Bradley (Studio 5000, L5X files) — full production support
- Siemens (TIA Portal) — full production support for any project that supports Siemens' Openness API (the vast majority of projects in production today); the Desktop Companion App uses Openness to export directly from TIA Portal, no manual export step
- CODESYS V3 — full support via the CODESYS .export flow, covering first-party CODESYS environments and OEM platforms built on CODESYS V3 such as WAGO, ABB, and Bosch Rexroth
- More vendors available on request

## Live plant data

The Desktop Companion App streams live tag values and tag history straight from the running controller — not just static analysis of the project file. This grounds troubleshooting in what's actually happening on the line right now.

## Supported standards

- PackML
- ISA-88

## Pricing

Starts at $128 per PLC project per month, annual billing (one license = one PLC project file from one physical controller).

Plans, billed annually: Starter $128/project/mo (1–10 projects), Growth $120/project/mo (11–30), Plant $112/project/mo (31–80), Enterprise (80+, custom pricing).
Monthly billing is also available (Starter $160, Growth $150, Plant $140 per project per month); annual billing saves 20%.
Pricing steps down automatically and retroactively as project count crosses each tier threshold.
Unlimited collaborators on every plan — pricing is per controller, never per seat.
Each project includes an estimated 800 AI queries per month, pooled across the whole team.
Free trial: all platform features, unlimited user invitations, up to 3 projects under fair usage, no credit card required.

## What is automation context?

Data context tells AI what the plant is. Automation context tells it what the plant does, and why. It is the full set of context an AI needs to reason about industrial control systems: the control logic itself, its version history, the line around it, and the standards and conventions it is held to. PLCs.ai builds this layer from the customer's own projects. Details: /automation-context

## Downtime facts and sources (from our free downtime calculator, /downtime-calculator)

- Unplanned downtime costs $36,000 to $2.3M per hour in manufacturing (Siemens True Cost of Downtime report, 2024)
- Average large plant: 25 unplanned stops and 27 downtime hours per month (Siemens, True Cost of Downtime, 2024), about 65 minutes per stop. Siemens separately reports 81 minutes on average to restart production.
- IEC 60050-192 and MIL-HDBK-472 split repair time into fault localization, diagnosis, correction and check-out; finding the fault is often the larger share. They do not assign fixed percentages.
- Unplanned downtime costs the process industries almost 5% of annual production (ARC Advisory Group, 2016)
- Loaded labour cost averages $46.60 per hour worked across US private industry (US BLS, ECEC, March 2026)
- PLCs.ai attacks the fault-localization and diagnosis share of repair time

## Manufacturing skills

Slash-command skills the agent uses to analyze PLC projects:

- **/cycle-time**: Detects waiting times and cycle-time bottlenecks in ladder logic or program logic
- **/dead-code**: Finds disabled, bypassed, or unreachable code
- **/interlocks**: Audits safety interlocks, bypasses, reset logic, and stored-energy de-energization
- **/missing-handshake**: Finds missing or broken handshake signals between sequenced stations
- **/race**: Finds race conditions, multiple writers, and scan-order bugs across tasks
- **/signal-trace**: Locates a tag and traces every read and write site across a project

Plain-English definitions of each concept (what is a cycle-time bottleneck, dead code, interlock audit, missing handshake, race condition, signal trace): /learn

## How PLCs.ai differs from ChatGPT, Claude, or any general-purpose chatbot

Pasting a PLC code fragment into a general-purpose AI chat tool (ChatGPT, Claude, or similar) loses cross-file context (tag aliases, UDTs, call trees), has no concept of a production line made up of multiple connected PLCs, has no PLC-specific skills, has no simulate-and-review step before generated code is applied to a live controller, and has no memory of an organization's own standards from one session to the next.

PLCs.ai: ingests the full project file and reasons across the whole project; with 3D Line Context, reasons across every PLC on a production line (even mixed Allen-Bradley/Siemens) as a single surface; reviews and simulates any generated code before it is saved; and runs every query against the organization's own knowledge base of approved code examples, standards, specs, and reference documents.

Full comparison: /compare/plcs-ai-vs-chatgpt

## Frequently asked questions

- **Can't I just paste my PLC code into ChatGPT or Claude?** You can, but a pasted fragment loses cross-file context, line context, safety review, and your organization's standards. PLCs.ai ingests the whole project and keeps all of that.
- **Is my PLC code used to train AI models?** No. Customer data is never used to train AI models, for any customer, on any plan.
- **What Studio 5000 files does PLCs.ai support?** L5X project exports from Studio 5000 / RSLogix 5000, legacy or modern, documented or not.
- **What TIA Portal projects does PLCs.ai support?** Any TIA Portal project that supports Siemens' Openness API — the Desktop Companion App exports directly from TIA Portal with no manual step.
- **Does PLCs.ai support CODESYS?** Yes. CODESYS V3 projects are supported via the standard .export flow, including OEM platforms built on CODESYS V3 such as WAGO, ABB, and Bosch Rexroth. Details: /blog/plcs-ai-supports-codesys-v3
- **Can I mix Studio 5000 and TIA Portal PLCs on the same line?** Yes. 3D Line Context reasons across every PLC on a line, including mixed Allen-Bradley and Siemens lines, as a single surface.
- **How much does PLCs.ai cost?** From $128 per PLC project per month billed annually ($112 at the highest volume tier), with unlimited collaborators and no seat fees.
- **What counts as a PLC project?** One license covers one PLC project file from one physical PLC controller.

## Integrations and enterprise connectivity

- Version control: GitHub, GitLab, Bitbucket, Copia, and octoplant — all five live today, connected read-write
- REST API: interpret, generate (reviewable proposal), multi-turn troubleshooting conversations, async analysis jobs, PLC/PDF exports — docs at /developer
- Embeddable iframe: short-lived, project-scoped, read-only tokens
- MCP server (`plcsai-mcp-server`): lets Claude Desktop, Claude Code, Cursor, Codex, or any MCP-compatible agent query projects, live tag values, and analyses directly — docs at /developer/mcp

## Security

- Data isolated per customer — no co-mingling between organisations
- Customer data is never used to train AI models
- AES-256 encryption at rest, TLS 1.2+ in transit
- OAuth 2.0, MFA/TOTP, SSO (Enterprise)
- SOC 2 compliance available for Enterprise plans
- Deleted data is soft-deleted immediately and permanently purged within 90 days

## Company

PLCs.ai was founded in 2025 and is based in Tel Aviv and London.
Founders: Zohar Kantor (Co-founder & CEO) · Noam Weisman (Co-founder & CTPO).
Raised a $4M seed round (announced February 2026) from Avila VC, Ocean Azul Partners, Sugar Free Capital, and ICI Fund.
Contact: security@plcs.ai · privacy@plcs.ai · sales@plcs.ai
Profiles: https://www.linkedin.com/company/plcs-ai · https://www.automate.org/companies/plcs-ai

## Pages

- / — Homepage: overview, platform pillars, problem statement, supported vendors
- /platform — Detailed platform capabilities and how it works
- /plc-troubleshooting — AI PLC troubleshooting: how PLCs.ai traces faults through the full project and confirms root causes against live controller values
- /pricing — Plan comparison, feature table, FAQ
- /automation-context — What automation context is and how PLCs.ai builds the intelligence layer from your production context
- /packml — PackML in the AI Era (white paper, free PDF): how standards-aware AI recognizes PackML, scores compliance, guides remediation and troubleshoots PackML machines and lines
- /downtime-calculator — Free calculator pricing the fault-finding share of unplanned downtime; every default cites a published source
- /data — Full data handling, security, and compliance documentation
- /vendors/studio-5000 — Rockwell Automation / Allen-Bradley Studio 5000 capabilities and FAQ
- /vendors/tia-portal — Siemens TIA Portal capabilities and FAQ
- /compare/plcs-ai-vs-chatgpt — How PLCs.ai differs from pasting PLC files into a general-purpose chatbot
- /learn — Glossary of PLC concepts (cycle time, dead code, interlocks, handshakes, race conditions, signal tracing) and the skills that detect them
- /blog — Product updates, engineering deep-dives, and company news

## Key articles

- /blog/ai-for-plc-programming — Can you use AI for PLC programming? Answers: which AI is better for PLC programming, is AI replacing PLC programmers, are PLCs becoming obsolete
- /blog/best-ai-for-plc-programming — The best AI for PLC programming, compared honestly: general chatbots vs Rockwell FactoryTalk Design Studio Copilot vs Siemens Industrial Copilot vs PLCs.ai, across 10 criteria including line context, live plant data, mobile access, and enterprise readiness
- /blog/can-ai-read-ladder-logic — Can AI read ladder logic? What works on real projects
- /blog/cost-of-unplanned-downtime-manufacturing — What unplanned downtime really costs in 2026: every figure sourced (Siemens 2024, ARC, US BLS, MIL-HDBK-472), plus how to price your own downtime hour
- /blog/plc-troubleshooting-guide — PLC troubleshooting: the complete eight-step field method, from status LEDs and fault codes to verification against live tag values, plus the most common PLC failure classes
- /blog/packml-explained — PackML explained: the 17-state model (with state numbers), Held vs Suspended, unit modes, PackTags (Command, Status, Admin), ISA-88 program structure, a compliance checklist, and where AI fits
- /blog/plc-fault-codes-hmi-alarms-root-cause — PLC fault codes and HMI alarms: how to trace an alarm to its root cause (alarm text, alarm bit, the rung that sets it, the false interlock, the live value, the physical device), plus which AI tools help at each step
- /blog/how-to-reduce-mttr-manufacturing — How to reduce MTTR in manufacturing: the repair-time anatomy (diagnosis is 60-70% of the clock), benchmarks, and the five levers that shrink fault localization
- /blog/real-time-plc-data-troubleshooting-multiplier — Runtime (real-time) PLC data multiplies AI troubleshooting: live tag values streamed from the running controller confirm or rule out root causes in seconds instead of a walk to the panel
- /blog/plcs-ai-vs-siemens-industrial-copilot — PLCs.ai vs Siemens Industrial Copilot: which one do you need
- /blog/how-to-document-plc-code — How to document PLC code automatically
- /blog/best-ai-for-industrial-automation — What is the best AI for industrial automation in 2026: industrial AI compared by the problem it solves and the context it understands (operations/data AI, asset health, machine vision, engineering copilots, controls intelligence)
- /blog/plc-documentation-software — PLC documentation software: 9 tools compared (2026), from vendor add-ins that print what is in the project, to AI that infers meaning from undocumented code, to context-aware platforms that generate documentation from the whole automation system; answers "is there an AI tool that documents PLC code automatically"
- /blog/plc-tribal-knowledge-retiring-engineers — What happens when your best PLC engineer retires: the tribal knowledge problem
- /blog/what-is-plc-ai — What is PLC AI? Definition, what AI can already do with a PLC, the five types of PLC AI tools on the market, and where PLCs.ai sits among them
- /blog/three-generations-of-ai-for-plc-work — The three generations of AI for PLC work: chatbots that know PLC concepts, file-aware AI that understands your project, and live-plant AI that sees what the line is doing right now
- /blog/mixed-vendor-plc-troubleshooting — Why industrial AI must understand the entire production line, not just one PLC: tracing a fault across an Allen-Bradley/Siemens handshake, and what mixed-vendor line context means
- /blog/manufacturing-ai-adoption-journey — The manufacturing AI adoption journey: from one engineer, one PLC, and one AI model to governed enterprise-wide access, and what infrastructure each stage needs
- /blog/plcs-ai-supports-codesys-v3 — PLCs.ai now supports CODESYS V3 projects across first-party CODESYS environments and OEM platforms like WAGO, ABB, and Bosch Rexroth, and why a .export with no tag table required real re-engineering
- /blog/automation-context-layer — Industrial AI has a context problem, not an intelligence problem: why models need an automation context layer (control logic, version history, the line around it, standards)
- /blog/give-an-llm-more-plc-context — Why not just give an LLM more PLC context? Where the more-context approach runs out: who curates it, how conflicts resolve, how it stays current, what happens when the AI is wrong
- /blog/plcs-ai-vs-chatgpt — PLCs.ai vs pasting your PLC files into ChatGPT: whole-project reasoning, simulate-before-save, and a guarantee your code never trains someone else's model
- /blog/ai-for-studio-5000-allen-bradley-plcs — AI for Studio 5000 and Allen-Bradley PLCs: explain, troubleshoot, generate, and document, grounded in the actual L5X project
- /blog/desktop-companion-app-tia-portal-export — Why TIA Portal needs a Desktop Companion App and Studio 5000 does not: S7 platform support and the Openness export path
- /blog/six-things-every-plc-engineer-runs-into — Six things every PLC engineer runs into (cycle-time bottlenecks, dead code, missing interlocks, broken handshakes, race conditions, signal tracing) and how PLCs.ai finds them automatically
- /developer — REST API documentation (also at developer.plcs.ai)
- /developer/mcp — MCP server documentation for connecting AI agents to PLC projects
- /status — Live system status and uptime (also at status.plcs.ai)

## Company news

- /blog/plcs-ai-raises-4m-seed-round — $4M seed round (February 2026) led by Avila VC, with Ocean Azul Partners, Sugar Free Capital, and ICI Fund
- /blog/brian-tessier-joins-as-advisor — Brian Tessier, former VP Innovation at Schneider Electric, joins PLCs.ai as an advisor (April 2026)
- /blog/changing-how-engineers-run-automation — Founding vision: if modifying a PLC project takes days, the problem is that no one fully understands the code
- /blog/automate-2026-chicago — PLCs.ai at Automate 2026 in Chicago, meeting engineers and plant managers without a booth
- /feed.xml — RSS feed of all blog posts