Story
September 14, 2026
As software-defined architectures continue reshaping industries from defense and aerospace to communications and advanced research, the future of testing will be defined not by owning more instruments but by giving engineers the ability to create exactly the instrument their application requires, precisely when they need it.
Artificial intelligence (AI) is transforming engineering in countless ways, from software development to chip design. Yet one of the most significant opportunities may lie in an area that receives far less attention: the tools engineers rely on to test, validate, and refine their ideas.
Over the past decade, engineering has undergone a fundamental shift toward software-defined architectures. Radios, electronic warfare (EW) systems, satellite payloads, instrumentation, and even vehicles increasingly derive their capabilities from software rather than fixed hardware. Yet while the systems themselves have become dramatically more flexible, the tools engineers use to test them have often remained surprisingly rigid.
Modern engineering has outgrown traditional test systems
As software-defined systems become the norm across aerospace, defense, embedded computing, communications, and advanced research, the demands placed on test equipment have changed dramatically. Engineers are no longer validating static hardware with predictable behaviors, but are instead developing highly configurable systems that evolve throughout the design cycle.
Yet many test platforms remain fundamentally fixed-function. While they continue to perform exceptionally well for established measurements, increasingly complex applications often require capabilities that simply don’t exist in commercial instruments. As a result engineers are developing flexible, software-defined systems using test equipment that wasn’t designed to keep pace.
From software-defined radios (SDRs) that reconfigure waveforms and protocols, to radar and EW systems that adapt to new threats in the field, modern systems are designed to evolve over time and with that, testing requirements evolve. As devices operate across a wider set of conditions and contend with an increasingly crowded and contested spectrum, these factors multiply the test cases engineers must validate.
The gap between standard instruments and custom solutions
When commercial instruments cannot perform the required measurement or control function, engineering teams have traditionally had only one option: build it themselves. That typically means significant effort in custom FPGA [field-programmable gate array] code before meaningful testing can even begin. While this approach offers tremendous flexibility, it also requires specialized expertise, lengthy validation, and development cycles measured in months rather than days.
For many organizations, particularly those working under aggressive development schedules or evolving mission requirements, that investment creates a significant bottleneck. Instead of creating the exact test configuration they need, engineers often compromise by adapting their work around the limitations of available instrumentation.
Historically, developing custom FPGA-based capabilities has required specialist expertise, not typically available to most hardware engineering teams or research labs. In addition to developing the VHDL or Verilog code itself, deploying the code to hardware introduces further challenges. While development boards are usually the lowest cost approach, interfacing to them and supporting them over time is difficult. User-programmable FPGA solutions from traditional test equipment vendors offer supported, easier to access tools.
Generative instrumentation changes the equation
The next evolution of instrumentation isn’t simply making existing tools faster or more capable, it’s making them adaptable.
Software-defined instrumentation built on reconfigurable hardware enables engineers to create application-specific measurement systems that evolve alongside the systems they’re testing. Rather than selecting the closest available instrument and working around its limitations, engineers can configure the capabilities they need for a specific application.
AI further reduces the barrier by simplifying how those instruments are created. Instead of manually implementing every function, engineers can describe the behavior they need while an agentic workflow generates, validates, and deploys the resulting instrument. The engineer remains firmly in control of defining requirements, evaluating results, and refining performance, with AI simply accelerating the path between concept and implementation.
Generative instrumentation enables fast, easy integration of application-specific capability and IP into test hardware. This allows engineers to not only optimize their test system to their device under test, but to do that over and over again as requirements evolve. GenInst Studio is an AI-enabled instrument creation platform that turns natural-language prompts into validated, customized test functionality that runs on Moku hardware. Through a guided specifications process, the tool gathers the user’s requirements and then generates instrument code. It validates the design by creating and running an extensive set of tests and iterates until all tests pass, enabling full transparency by providing the code and tests for the user to audit. The new custom instrument is then ready to deploy to the Moku FPGA, running in hardware, ready to connect to real-world signals or integrate into a larger test system. (Figure 1.)
[Figure 1 ǀ The AI-enabled GenInst Studio instrument-creation tool enables the user to deploy a new custom instrument to an FPGA, running in hardware, ready to connect to real-world signals or integrate into a larger test system. Liquid Instruments graphic.]
Faster test development enables faster innovation
Across defense, aerospace, and advanced research, development cycles continue to compress while system complexity increases. Modern engineering programs more and more rely on rapid iteration, whether developing new radar modes, communications protocols, embedded control systems, or advanced sensing technologies. The ability to quickly adapt testing to changing requirements becomes just as important as the system being developed.
Reducing the time required to create specialized test capabilities enables engineering teams to validate more ideas, iterate more frequently, and respond more quickly to evolving technical challenges. In many cases, accelerating the development of the test environment ultimately accelerates development of the product itself.
One common but powerful example is custom triggering. This could be used in EW to detect a specific signal of interest, or in a fault monitoring system to ensure safe operation. Triggering is time-sensitive and requires the deterministic, low-latency performance that only a hardware-based solution can provide. Generative instrumentation enables users to define their trigger conditions to exactly match their system or signals, matching the flexibility of software, but with the performance of hardware.
Another example is real-time signal processing, which can be used for applications like filtering and data reduction to measurement acceleration. User-defined digital signal processing algorithms are deployed in line with signal acquisition or generation and executed in real time, on every sample. As data rates increase and the amount of data being acquired or generated increases, this capacity becomes even more critical, as capturing and post-processing all that data becomes prohibitive. Averaging, filtering, or calculating a specific result in hardware along with the ability to refine and reconfigure as testing evolves has major potential for faster iteration and better results.
Generative AI: the next evolution of software-defined test
Software-defined systems have changed how engineers design products; because of this shift, test environments must evolve in parallel. Rather than relying solely on collections of fixed-function instruments, engineering teams need adaptable platforms capable of evolving alongside rapidly changing applications. Reconfigurable hardware, combined with intelligent software, makes that flexibility accessible to a far broader range of engineers than ever before.
AI is not here to replace engineering expertise. Instead, the goal is to remove much of the tedious complexity involved in creating specialized instrumentation and enable engineers to focus on solving technical problems rather than spending all that time building the tools needed to investigate them.
Liquid Instruments CEO Daniel Shaddock’s research focuses on precision measurements using advanced digital signal processing. He led Australia’s involvement in GRACE Follow-on, a satellite mission launched in 2018 to track the Earth’s water movement. Prior to this work Daniel was a Director’s Fellow at NASA’s Jet Propulsion Laboratory where he served as NASA’s Interferometer Architect for the LISA mission. He is a Fellow of the American Physical Society and was a co-author on the paper announcing the observation of gravitational waves, an achievement that was awarded the 2017 Nobel Prize in physics.
Liquid Instruments
