As engineered systems become more connected and software-defined, testing can no longer remain a final checkpoint before product release. Ritu Favre, president of Emerson’s Test & Measurement business group, argues that test and validation must become continuous parts of engineering workflows, supported by connected platforms and artificial intelligence, tells IEEE Spectrum.
Semiconductor development illustrates the challenge. Chiplet-based designs can combine components, interposers, and packaging from different suppliers. Each may function correctly independently, yet unexpected behavior can emerge when they operate together. Similar complexity appears in automobiles and aircraft, where software, sensors, electronics, and mechanical systems must constantly interact under changing real-world conditions.
Traditional testing methods often discover problems late in development, when failures are more expensive and difficult to diagnose. Integrating validation earlier allows engineers to identify problems while designs are easier to modify. Test data can then become a source of engineering insight instead of simply producing a final pass-or-fail result.
Connected test platforms play an important role by linking information across design, validation, and production. Rather than rebuilding separate workflows as requirements change, engineers can use adaptable platforms that scale with new technologies and make measurement data available across development stages.
AI could make these connected systems more valuable, but its effectiveness depends on structured, traceable engineering data. With sufficient context, AI can analyze interactions among multiple systems, identify unusual behavior, and help engineers locate the root causes of failures. It is intended to support rather than replace engineering judgment.
As product complexity increases, verification is becoming as important as innovation itself. Combining continuous testing, connected data, and AI could help engineering teams understand system behavior earlier, reduce late-stage problems, and develop products that perform reliably under real-world conditions.
