A Deep Dive Into Nigel: NI’s New AI Advisor

According to Moore’s Law, which was first observed in 1965, the number of transistors in semiconductor chips doubles every two years. This pales in comparison to the rapid explosion of Artificial Intelligence (AI) tools, which have been experiencing exponential growth at 2-3 times that rate! In a recent webinar, engineers from Emerson/NI gathered to discuss the role that AI agents play in test and measurement applications—focusing on their newest AI Advisor, Nigel. This post will dive into the webinar’s discussion, highlighting the key takeaways and outlining the thought-provoking Q&A.

How Do Test Engineers Benefit From AI-Driven Tools?

AI agents are systems that use existing tools to develop workflows and carry out activities. The capabilities of these agents go beyond natural language processing, branching into making decisions, solving problems, or interacting with the outside world.

In the webinar, titled “Embracing AI in Test—LabVIEW, Nigel™ AI Advisor, and Beyond,” NI Business and Technology Fellow Charles Schroeder joined Taylor Riche, the Director of Technology and Innovation Software for Emerson/NI. They began their discussion by positing that AI is an incredibly unique tool, as it is the first created that can also create other tools. Schroeder brought up two main categories in which AI can greatly benefit test engineers—productivity and insight. As test engineers are faced with increasingly complex testing environments, AI tools can help boost productivity by determining how to get a quality product to market faster.

When it comes to measurement data insights, many test engineers are unsure of how to manage all of the data they receive. Schroeder addressed this issue by highlighting four types of data that are the most important for test and measurement engineers:

1) Information About the DUT (Device-Under-Test)
What is the product supposed to do? How do we guarantee that the product meets all necessary requirements?

2) Information About the Test System
What instruments are being used? What are their accuracies?

3) Information About the Software Written to Make Measurements
How is it written? What algorithms are in them?

4) The Measurement Data Itself
Is it linked to the context in which it was taken?

Schroeder asserted that if these four pieces of information are well understood by the test engineer, AI can do amazing things!

Overview of the Nigel™ AI Advisor

Unlike other AI platforms that are more general-purpose, NI’s Nigel AI Advisor is specifically tailored for the creation of testing applications. Nigel is trained on decades of reliable data—offering insights, automation, and contextual assistance presented in tables and graphs.

Trained for both LabVIEW and TestStand, Nigel operates as a plug-in within LabVIEW with a Chat UI that is local to the desktop app. Both the local and the cloud services are encrypted, and the cloud allows Nigel access to all of NI’s test and measurement knowledge (built on over 50 years of experience). There is currently a usage limit of 500 requests a day, or 20 VI descriptions. However, this is subject to change in the future.

To access Nigel, you will need an updated subscription to LabVIEW Professional. The “Project Save” feature for LabVIEW can be found in the project settings for LabVIEW 2025 Q3, and it allows you to use the latest editor without updating all of your code.

The general architecture of Nigel is built on top of OpenAI’s GPT models, and it is deployed to Emerson’s secure cloud operations platform. One of the biggest advantages of Nigel is that it doesn’t just use broad information, it knows about your specific application and code.

For top-of-the-line security, Nigel is hosted on Microsoft Azure. In fact, security was a top priority for NI when creating Nigel. NI guarantees that your data is not stored by Nigel nor used to train models, it is stored locally on your device only. In addition, all network traffic is encrypted (TLS).

Moving forward, some exciting improvements that are being developed for Nigel include expansion into FlexLogger and Instrument Studio, a focus on hardware and performing measurements, and the ability to deploy to a personal cloud.

Practical Workflow Demonstration with Nigel

To demonstrate the capabilities of Nigel during the webinar, NI Solutions Marketer David Prida set up a practical application example. In the example scenario, he was acting as an engineer who was tasked to measure output ripple from a buck converter.

For the first prompt, he asked Nigel to explain measuring output ripple from a buck converter. As a result, Nigel responded with an in-depth answer filled with well organized information. It explained what output ripple is and how to measure it, providing a step-by-step guide as well as practical tips and reference images with further reading suggestions.

The second prompt asked Nigel for help with recommending a PXI scope to assist with that application. Nigel has an extensive knowledge of NI hardware, and provided practical recommendations alongside a list of key selection criteria in response to the query.

The third and final prompt in the example scenario asked Nigel for a LabVIEW example of performing that measurement testing the knowledge of NI software in addition to hardware. In response, Nigel was able to give a specific function name, describe the VI, and provide a diagram. To take things a step further, David asked Nigel how to augment/modify the code for vpp measurement.

This demonstration showcased Nigel’s vast knowledge of NI’s hardware and software, highlighting its ability to enhance testing workflow productivity!

FEATURED QUESTIONS AND ANSWERS

Question: “Can Nigel access my own VIs and codes to suggest improvements?”
Answer: Any normal VI in your project can be described by Nigel. It currently only uses a single VI if you prompt with “describe vi”, but the ability to use more context from your project is a future feature.

Question: “Can Nigel write Python code, or code for an instrument driver?”
Answer: Nigel (in LabVIEW 2025 Q3) can not write code for those programs yet, but that capability is in the works.

Question: “Is Nigel coded for other languages?”
Answer: Yes! Nigel currently supports 95 languages.

Question: “How is Nigel trained, if not with customer data?”
Answer: Nigel is not “trained”, it is “taught” / “augmented”. Trained models may be used in the future with NI’s data.

Question: “Can Nigel help in code with other vendor hardware?”
Answer: Base models do have vast amounts of knowledge of the world in general, but if you are asking for specifics about things outside of NI’s walls, it may be inaccurate.

Question: “What is the advantage of Nigel over an LLM such as ChatGPT or Gemini?”
Answer: First, the types of answers that Nigel gives. Nigel focuses on the specific domain instead of trying to solve things generally. Second, the facts that have been used for Nigel are specific to test and measurement. As Nigel continues to get smarter, it will get more insights. NI is also holding themselves to a higher standard for answer quality than other LLMs

Question: “Can Nigel produce documentation?”
Answer: Yes! Nigel can produce documentation for code that doesn’t have any.