Control Hub
AI features
Reduce debugging time with AI that works directly with your instruments.
Spend less time configuring decoders, building triggers and repeating measurements. Keep control of your data with local Auto-decode and your choice of a local or hosted assistant model.
01 / Connect
MCP server
Investigate a problem without copying settings and results between tools.
Control Hub’s Model Context Protocol (MCP) server connects compatible AI applications to your bench. Discover connected instruments, read their settings and measurements, and carry out acquisition tasks from the client you already use.
- Discover instruments and their available controls.
- Set channels, acquisition settings and supported triggers.
- Start or stop a capture and retrieve its results.
- Read measurements and decoded protocol data.
The available controls follow the connected instrument. Use the same connection for Modular Bench hardware and supported third-party equipment.
Supported instrumentsOne connection to your bench
- ClientYour AI application
Your choice of model and workflow
- InterfaceMCP server
Instrument tools and measurement results
- ControlControl Hub
Connected instruments and their controls
02 / Decode
Auto-decode
Get readable data without manually working out the protocol and settings.
Auto-decode uses a small local AI model to identify the protocol, assign channel roles and work out decoder settings from a capture. Select the channels that belong to a bus; Control Hub sets up the decoder and displays the decoded data.
- Recognize protocols such as I²C, SPI, UART and CAN.
- Infer clock and data inputs, signal rates and protocol settings.
- Use captured logic signals or digital buses measured by the oscilloscope.
- Repeat for another bus without replacing existing decoders.
When more than one configuration fits, choose from the suggested matches. You can adjust the resulting decoder settings at any time.
Runs locally with Control Hub. No cloud model or API key required.
Protocol decoding03 / Trigger
Vibe Trigger
Catch the event that explains the fault, without building every trigger step by hand.
Build a custom trigger sequence in plain language. Vibe Trigger turns your description into steps with channel assignments, trigger conditions and timeouts, ready to review in Control Hub.
- Combine edges, pulse conditions and logic patterns.
- Specify which event comes first and what must follow.
- Add a timeout to restart when the expected event does not arrive.
- Refine the request or edit the sequence before arming.
The sequence uses the trigger types and step limits supported by the selected instrument. Once armed, the instrument evaluates the trigger conditions; the AI model is not in the trigger timing path.
Hardware trigger typesSequence examples
After a rising edge on D0, trigger on a high pulse shorter than 2 µs on D1. Restart if it takes more than 1 ms.
- 01Rising edgeD0
Advance to step 2
- 02High pulse < 2 µsD1
Trigger capture · 1 ms timeout
The final step triggers the capture. A timeout returns the sequence to its first step.
04 / Control
AI assistant
Spend less time repeating setup and measurement steps.
Bring your own API key and ask the assistant to configure an instrument, inspect a capture, find a signal or compare decoder settings. It operates Control Hub’s controls and shows each action as it works, followed by a summary of the result.
- Adjust instrument settings and acquisition controls.
- Search channels and set up protocol decoding.
- Read measurements and investigate captured signals.
- Run repeated checks and measurement workflows.
Open ASK or press Ctrl + K. The instruments stay visible while you chat. Follow-up requests keep the conversation context, and you can stop a running request from the assistant panel.
Example requests
Find CAN at 500 kbit/s on the logic analyser and show the decoded frames.
Measure the peak-to-peak voltage and frequency of each scope channel.
Set the power supply to 5 V with a 100 mA current limit. Leave the output off.
Actions appear in the assistant panel and the instrument controls.
Your model connection
Keep control of where your data goes.
In Control Hub’s Assistant settings, enter an OpenAI-compatible model endpoint, select the model and add your API key if the service requires one. Use a model with tool calling; image input is needed when asking it to inspect a screenshot.
Choose a model running on your own computer or network to process assistant requests locally. Choose a hosted model when you want to use that service; prompts and relevant instrument context are sent to that provider, and usage is billed through your own account. Auto-decode runs locally in either setup.
- 01
Connect
Enter the model endpoint and your API key.
- 02
Select
Check the connection and choose a model.
- 03
Ask
Open the assistant and describe the task.