The Acute Need for Software Automation in Modern Test Labs

It was not that long ago that a pair of headphones mainly consisted of two speakers and a wire. Then came the addition of a battery and wireless radio. Then microphones too, new controls, noise canceling, significant onboard processing and more… People now walk around with quite powerful computers in our ears, and consumers expect them to be lite-weight, reliable, powerful and affordable!

It’s a tall order for any design engineering team to match the pace of innovation expected in a competitive market. But it’s arguably tougher for the test engineer as these groups rarely get the same level of project time or resource investment as their Research and Development counterparts. Validation (and manufacturing) groups across industry are feeling increasingly stretched as the gap between their capacity and their project requirements is growing. This is forcing test leaders to accept they must change their methodology or risk product failure or market slowness.

One big change in methodology is the widespread adoption of greater software automation across validation labs in almost every test they do. This is for three main reasons.

1. Test coverage: Conducting larger sweeps of test cases and vectors provide a more complete characterization and understating of a device under test (DUT). This leaves less room for errors in identifying potential faults. This level of testing would take forever with a manual or semi-automated set-up, software orchestration allows for much higher test coverage in much less time.
2. Repeatability: There is nothing more frustrating than repeating a test and finding a different result due to variances in test set-up rather than DUT behavior. Centralized software automation of the instrumentation provides the ability to save a complete lab set up, so that in the future there can be a toggle between complex test configurations involving multiple instruments at the click of a button.
3. Data access: Measurement data has been recognized as a new source of value for many product companies to derive wider product and operational performance trends – but it has no value if it never gets beyond the instrument. Software automation architectures that use modular instruments require the engineer to have immediate access to the measurements needed – while simultaneously streaming large contextual data sets into centralized storage for use by other people and artificial intelligence (AI) models.

New technology is democratizing Software Automation

The traditional challenge to software automation has been the high barrier to entry in writing the custom code required for each unique measurement that engineers may want. However, this is changing, first because of more complete application software that provides configuration-based workflows for complex measurements and second because generative AI mixed with graphical coding paradigms have lowered the barrier to entry for many engineers.

Application Software: What’s faster than vibe-coding an application – using an application that already exists! Millions of hours of development have collectively been put into engineering tools for test and measurement applications. Take National Instruments (NI) InstrumentStudio the companion software to PXI that is used to configure and visualize measurements. When referencing multiple oscilloscopes such as the NI PXIe-5108 within a single chassis they automatically synchronize so they can be treated as though they were a single, high channel count instrument. The software infrastructure works in the driver to enable this is entirely abstracted from the user. To unnecessarily recreate this wheel every time degrades standardization benefits and introduces both time and risk.

As the user considers the workflow of tasks that together represent the role of a test engineer, more and more of them have application software targeted specifically to solve each problem (Figure 1). NI TestStand is the world's leading test management and sequencing tool. NI DIAdem has established itself as the go-to for data analysis and report generation for large data-sets. While NI SystemLink delivers remote management capabilities, saving teams from writing and maintaining web-based monitoring tools. Every time engineers opt for a test optimized tool rather than creating something new, or compromising with a generic alternative, they squeeze a little more productivity out of their already packed schedule.

Figure 1: NI LabVIEW+ Suite: 788509-35. (Image source: NI)

Generative AI and Graphical Coding Paradigms: What’s faster than writing an application: writing an application with the assistance of generative AI. However, as far as this technology promises a lot, it also should be used with caution due to the risk and quality issues it can introduce.

When code is generated (human or AI), the designer is still accountable for the correct operation and stability of it. That includes verifying measurements, identifying edge cases, and ensuring that results can be trusted. If an issue is missed, the consequences are tied to the outcome of the test, not the tool used to create it.

Measurement applications contain some unique properties that differentiate them from software development in other industries such as web, database or financial applications. The big difference is that they interact with the real world – streaming large quantities of mixed signal data into a program creates needs for parallelism, time domain knowledge and often varied hardware target deployment, event interruptions and an operator UI.

Tools that make behavior visible—showing how data flows, how processes interact, and how code is deployed—make it easier to understand and validate those systems.

Graphical code is not inherently easier to understand for every type of application – but it does make measurement application development more straightforward, as it represents the operation of these systems really well.

The distinction is similar to how circuits are approached. A circuit can be represented as a netlist, a complete description of components and connections, but when diagnosing behavior, engineers rely on schematics because they make relationships and dependencies visible. The choice of a development environment is not just for accurate representation, but for effective understanding too.

LabVIEW applies the same principle to test systems. It provides a visual model that makes parallel processes explicit, allowing engineers to trace data flow and identify issues as they occur rather than stepping through execution sequentially. The user can easily visualize the operation of two or more processes which have important dependencies on each other. 

Generative code generation will come to LabVIEW in the Summer of 2026 through the latest release of Nigel AI. But more than this, LabVIEW has always been an open environment, and so the answer to tool choice lies in combining tools rather than separating them.

Using a language/AI tool of choice to vibe code tasks, algorithms or processes that do not have complexities related to measurement applications could accelerate development as long as the code made is trusted. This is similar to how in LabVIEW engineers have always imported or referenced IP using DLLs or native integration nodes. Then by integrating this into the LabVIEW environment, the benefits of application structure, hardware/data interaction, UI, target deployment, etc. can be enjoyed. This is a “best of both worlds scenario”.

Test Engineering is Greater Task than Code Generation

The scope of what test engineers are responsible for has outgrown any single development environment. Today, the LabVIEW+ Suite provides tools that span planning, simulation, bring-up, development, de-bugging, data visualization and more. As the expectation on test professionals grows, they need dedicated tools that allow them to work at higher levels of abstraction across more of the tasks they face.

Here, AI plays an even greater role in driving productivity for the individual. This could be in test plan generation, asset management, process optimization, data management, root cause analysis or one of any number of important tasks involved with cradle-to-grave test system design. To be effective, the AI tool used must understand the whole system, have contextual awareness between tasks, have good understanding of test methodology best practice, and maintain checks that “keep an engineer in the loop”.

Tools such as NI TestStand, NI FlexLogger, and NI InstrumentStudio address different parts of the workflow, but engineers move between them without losing contextual awareness of the task they are working on. The boundaries between these tools will continue to blur with Nigel AI acting as a bridge between the unique functionality they offer.

The move from a user orchestrated workflow to an AI orchestrated workflow (Figures 2 and 3) has greater potential to drive productivity in test engineering than over-optimizing test IP development by language.

Figure 2: Transform Test Systems with NI’s Open, Flexible Platform: NI PXI and NI CompactDAQ. (Image source: NI)

Figure 3: Nigel AI serves as a bridge between NI’s other testing platforms. (Image source: NI)

Summary

Hardware choice is essential to being ready for an AI enabled test world. Combining testing tools to aid in AI code generation will increase productivity, functionality, and many other system tasks. For more information on the future of software testing, read How Software is Defining the Future of Test: A Q&A with Rudy Sengupta from NI.