Software Quality Assurance Analysts and Testers

• Manual code tracing1956–1979

  Manual code tracing and desk-checking (pre-disciplinary era)

  In the earliest computing era, testing consisted of tracing through code by hand, comparing against expected outputs, or submitting batch jobs and reading printout results. There were no testing tools, no frameworks, and no methodology. Gerald Weinberg's IBM team in the late 1950s represented the most advanced practice of the era: a group of people whose explicit job was to attempt to break software before it went live. The 1968 NATO Software Engineering Conference in Garmisch, Germany formally named the "software crisis" and identified inadequate testing as a root cause of catastrophic cost overruns in government computing projects. Testing remained informal and ad hoc for most organizations through the 1970s.

• Structured testing methodology1979–1991

  Structured testing methodology (Myers 1979, IEEE Std 730, black-box / white-box discipline)

  The publication of Glenford Myers's "The Art of Software Testing" (1979) and the simultaneous release of IEEE Std 730 (Software Quality Assurance Plans) gave the profession its intellectual foundation. Myers introduced the distinction between black-box testing (testing behavior from the outside, without knowledge of internal structure) and white-box testing (testing code paths and logic structures). Equivalence partitioning and boundary value analysis, formalized in this era, remain standard practice today. The discipline acquired its first academic treatment, its first professional vocabulary, and the beginnings of organizational independence from development teams. ANSI/IEEE 829-1983 established standards for software test documentation, giving QA managers the paperwork scaffolding to run formal test cycles.

• Commercial GUI test automation1991–2004

  Commercial GUI test automation (Mercury WinRunner 1991, QuickTest Professional 1998)

  Mercury Interactive launched WinRunner in 1991, introducing record-and-playback GUI automation for Windows applications. For the first time, a tester could record a sequence of user interactions and replay it automatically against a new build, checking for regressions without re-executing each step by hand. QuickTest Professional (QTP), introduced by Mercury in 1998 and using VBScript for test logic, became the dominant enterprise automation platform of the early 2000s. These tools created the first real skill divide within the profession: testers who could script in TSL or VBScript commanded materially higher salaries than those who could not. HP acquired Mercury Interactive in 2006 for $4.5 billion, rebranding QTP as HP Unified Functional Testing. The commercial automation era defined the "QA Engineer" as distinct from the manual tester and set the trajectory toward the SDET role.

  Effect on the work

  Commercial automation tools reduced the labor required to execute regression test suites and enabled QA teams to maintain coverage as codebases grew. However, tool fragility (record-playback scripts break when UI elements change) meant maintenance overhead was high, and the era created as many specialized jobs maintaining test scripts as it eliminated in manual testing.

• Open-source web testing2004–2014

  Open-source web testing (Selenium 2004, JIRA 2002, Agile/continuous integration era)

  Selenium, released in 2004 by ThoughtWorks engineer Jason Huggins as an internal tool for testing web applications, became the dominant web testing framework globally within a decade of its open-source release. Unlike Mercury's proprietary tools, Selenium ran in real browsers and was free, which democratized web automation. Simultaneously, Agile methodologies shifted QA from a separate "test phase at the end" to a continuous responsibility integrated into each sprint. The JIRA issue tracker (Atlassian, 2002) became the universal defect management platform, and continuous integration tools (Hudson / Jenkins, ~2005) began running automated test suites on every code commit. The "shift-left" principle, testing earlier in the development cycle, became the professional norm. ISTQB, founded in Edinburgh in 2002 and launching its Foundation syllabus in 2005, gave the profession its first globally recognized certification structure.

  Effect on the work

  Selenium and CI tools dramatically raised the productivity ceiling for automation engineers. A single SDET maintaining a Selenium suite could provide coverage that previously required a large manual testing team. This accelerated a bifurcation of the profession: automation-capable testers became more valuable, manual-only testers came under wage pressure.

• Cloud testing platforms2014–2022

  Cloud testing platforms, mobile device farms, API testing (BrowserStack 2011, Postman 2014, Docker-based CI)

  Mobile computing created an entirely new testing problem: applications needed to work across hundreds of device-OS-browser combinations. BrowserStack (founded 2011), Sauce Labs (2008), and similar real-device cloud platforms let QA engineers run tests against physical devices without maintaining local device labs. Postman (2014) formalized API testing as a first-class QA discipline alongside UI testing. Docker-based containerized test environments, Jenkins pipelines with parallelized test execution, and cloud-based test infrastructure turned CI/CD from a development practice into a quality engineering discipline. "Test engineer" became a well-compensated senior engineering title rather than an entry-level checking role at leading technology companies.

• AI-native test automation2017–present

  AI-native test automation (Mabl 2017, QA Wolf, Testim, Diffblue Cover)

  Mabl, launched in 2017 and marketed as "AI-native" from its founding, represents the first generation of test platforms where machine learning is the primary engine rather than a bolt-on feature. Mabl's auto-healing automatically updates tests when UI elements change, addressing the chronic brittleness of record-and-playback systems. QA Wolf (2019) and Testim (Tricentis) generate production-grade Playwright and Appium test code from natural language prompts, promising 80% automated E2E coverage in four months. Diffblue Cover (2018) generates entire Java unit test suites autonomously, achieving 81% average line coverage versus 32% for a developer using a coding agent. These tools automate the work that occupied entry-level manual QA roles and much of the repetitive scripting work of mid-level automation engineers. The QA profession's response is bifurcating: roles focused on test execution and script maintenance face real displacement, while roles focused on test strategy, risk coverage design, and oversight of AI-generated suites are growing in both headcount and compensation.

  Effect on the work

  Tricentis research (2025) found that customers achieved 68% faster test suite creation after adopting AI-based modules, and 85% reduction in manual sanity testing effort. Over 40% of application code written in 2025 was AI-generated, and 29% of developers rolled back releases due to AI errors, increasing rather than decreasing the strategic importance of quality engineering as an oversight function. Median time-to-fill for senior automation roles exceeded 90 days in North America in 2025.

What's changing in your day

Three parts of your work where AI is already doing real lifting, and what stays yours.