Modu AI

No setup. Anonymous by default.

Industrial Device AI Workspace

For serial Modbus/AT devices, Modu AI turns manuals and product catalog entries into executable device capabilities: registers, function codes, AT commands, and parameter ranges. Describe the target setup in natural language, then let the app connect to serial ports, read status, map tool calls, ask before real writes, and keep TX/RX records.

Download Modu AI Send Feedback

From device information to real protocol actions

Modu AI is not a generic chat shell or a PLC programming copilot. It is a desktop workflow for serial device configuration, combining document parsing, DeviceScheme generation, Modbus/AT tools, write confirmation, and serial-frame records.

Compile documents into device capabilities

Import Markdown/TXT/HTML, PDFs or images, pasted text, URLs, or Public Document catalog entries. The app extracts registers, function codes, AT commands, default serial parameters, and parameter limits, then generates a DeviceScheme that Chat can use.

Use natural language to drive real tools

After you bind a document in Chat, AI can call serial connection tools, Modbus FC01/02/03/04 reads, FC05/0F/06/10 writes, AT mode commands, raw AT commands, and LoRa query/set tools.

Confirm writes and inspect what happened

Read operations return directly. Writes to coils or registers, AT write commands, and LoRa parameter changes create confirmation cards first. After approval, the app keeps tool calls, low-level commands, TX/RX frames, and traces inspectable.

Turn device information into executable, confirmable, traceable configuration flows.

The desktop app stores documents, sessions, and runtime state locally. The built-in cloud AI and document parsing service work anonymously. PDFs and images are converted to Markdown first, then the app extracts evidence-backed capabilities and validates them locally to reduce invented parameters.

  • DeviceScheme
  • Modbus/AT
  • TX/RX Trace

Start in 3 Steps

Three steps from installation to your first controlled device action

  1. Install the desktop app

    Install the desktop app and use the built-in cloud AI immediately — no sign-in or API key required.

  2. Import docs or choose a product

    Import a local file, pasted text, URL, or Public Document catalog entry to generate executable device capabilities.

  3. Connect serial and confirm execution

    Describe the goal, let AI map Modbus/AT tools, run reads directly, and confirm writes before execution.

FAQ

Which device types are supported?

The current focus is serial Modbus RTU and AT-command devices such as serial gateways, IO modules, LoRa modules, and DTU/RTU hardware. Built-in tools cover serial connection, Modbus coil/discrete-input/holding-register/input-register reads and writes, plus AT and LoRa query/set workflows.

Do I need to configure an API key or sign in first?

No. The default mode uses the Modu AI anonymous cloud service, so you can start after installation. Signing in can help with account and quota management, and BYOK can use your own OpenAI-compatible endpoint, but neither is required to begin.

How can I import manuals, and what if parsing fails?

You can import Markdown, TXT, LOG, HTML, pasted text, URLs, PDFs, common image formats, or choose a product from Public Document. PDFs and images are converted to Markdown through the Modu document parsing service. DOCX is not imported directly yet, so convert it to PDF or text first. Failed analyses can be retried in Documents.

How are write actions protected?

Write-oriented actions such as Modbus coil/register writes, AT write commands, and LoRa parameter changes create confirmation cards. Confirmation tokens expire, and unconfirmed writes cannot silently continue.

Where are my data stored?

Document records, parsed text, sessions, runtime state, and serial-frame logs are stored in local app data. When you use the built-in AI or PDF/image parsing, the relevant requests go through Modu cloud services. If you switch to BYOK, AI requests use the OpenAI-compatible endpoint you configure.

What concrete problems can it help with?

Typical tasks include reading DI/DO/AI status, setting a Modbus slave address, changing baud rate or serial format, querying device IDs, signal strength, and network registration, configuring LoRa channel, speed, address, or key, and keeping low-level commands and responses available for troubleshooting.