As generative AI accelerates software development, enterprises are facing a less visible problem: the faster software is produced, the harder it becomes to determine whether it is actually ready for production. Tricentis is targeting that gap with three new AI technologies spanning autonomous application exploration, AI-agent evaluation and release-risk analysis—part of a broader push to make software quality engineering more adaptive to agentic AI.
The software industry has spent the past two years racing to put AI into the development process. Coding assistants can generate functions, agents can modify repositories and AI systems can automate parts of testing. But that acceleration creates a new bottleneck: how do engineering teams validate software when the systems themselves are increasingly probabilistic and autonomous?
Tricentis is positioning quality engineering as part of the answer.
At its Tricentis Transform conference, the software testing company announced three technologies developed through Tricentis Labs, an innovation program designed to put emerging AI capabilities in the hands of customers and partners before they become fully commercialized products.
The releases—Tricentis Aida, Tricentis AgentScore and Tricentis Release Risk Intelligence—target different stages of the software development lifecycle. Together, they point to a broader shift in software testing from scripted validation toward AI-assisted and AI-driven quality decisions.
The timing is significant. Gartner has identified AI application testing as an emerging challenge because large language model applications can behave nondeterministically, making traditional software-testing approaches insufficient on their own. Its April 2026 guidance specifically addresses the adoption and scaling of AI application and agent testing.
From test scripts to autonomous exploration
Tricentis Aida is designed to explore web and Windows desktop applications autonomously. According to the company, it can identify defects and coverage gaps without an existing test suite, scripting or extensive setup.
That approach tackles a longstanding weakness in conventional test automation. Automated tests are powerful when teams know what needs to be tested, but building and maintaining those test cases can itself become a significant engineering workload.
An AI agent that can navigate an application, identify potentially problematic behavior and surface areas that deserve additional testing changes the economics of that process.
It also reflects a wider movement in software engineering toward agents that can perform multi-step tasks rather than simply generate code. McKinsey’s 2025 research found that 62% of organizations surveyed were at least experimenting with AI agents, while 23% said they were scaling an agentic AI system somewhere in the enterprise.
Aida is therefore less about replacing conventional testing than expanding what can be explored automatically.
Evaluating AI agents requires a different definition of quality
The more consequential innovation may be Tricentis AgentScore.
Traditional software tests often expect deterministic behavior: given the same input and environment, the application should produce an expected result. AI agents do not necessarily behave that way. Their outputs can vary, and the quality of an agent depends on factors such as task completion, consistency, safety and behavior across different real-world scenarios.
AgentScore is designed around that probabilistic environment. Tricentis says the technology observes AI agents operating in real-world workflows, recommends what should be measured and produces composite quality scores accompanied by recommendations to review, block or ship an agent.
That represents a fundamental change in quality engineering.
Rather than asking only, Did the software pass its test cases?, engineering teams increasingly need to ask, How reliably does this AI system perform its intended job under changing conditions?
That distinction is becoming important as enterprises deploy AI agents into customer service, software development, IT operations, finance and other workflows.
For enterprise technology leaders, an evaluation framework could become as important as the underlying agent itself. AI-generated code can be inspected. Autonomous agents require continuous observation of behavior.
Release management gets an AI layer
The third technology, Tricentis Release Risk Intelligence, moves further downstream.
It is designed to help quality leaders and release managers understand what has changed in a release, identify coverage gaps and prioritize potential risks. The system can then surface AI-powered actions intended to help teams investigate those risks.
That positions AI not merely as a testing tool but as a decision-support layer for software releases.
The distinction matters in large enterprises, where a release can involve hundreds of changes across applications, services, APIs and infrastructure. The challenge is often not finding another test to execute; it is determining which risks deserve attention before the release goes live.
Tricentis is also building these capabilities into a larger agentic quality engineering strategy. The company recently acquired Tabnine, bringing AI-powered software development capabilities into its broader platform and adding what Tricentis describes as an enterprise context layer alongside orchestration, governance and agent collaboration.
The competitive landscape is consequently expanding beyond traditional testing vendors. Microsoft, GitHub, Google, Amazon, OpenAI and specialized development platforms are pushing AI agents deeper into software engineering. Companies such as Selenium, BrowserStack, Sauce Labs and UiPath occupy adjacent portions of the testing, automation and application-quality ecosystem.
Tricentis’ differentiation is its attempt to connect AI development with quality engineering and release governance rather than treating testing as a separate stage.
That may become increasingly important as AI-generated software increases development velocity.
McKinsey’s research into AI-enabled software development found that more than 90% of surveyed software teams were using AI for development activities and reported saving an average of six hours per week. Its highest-performing organizations reported improvements of 16% to 30% in productivity, time to market and customer experience, alongside 31% to 45% improvements in software quality.
Those numbers come from McKinsey research rather than independent validation of any Tricentis product. But they highlight the underlying business case: when AI increases software production, quality systems have to scale with it.
For enterprise teams, that means evaluating AI quality platforms on more than automation claims. Integration with existing CI/CD pipelines, application observability, governance controls, data security, model evaluation and human review will become critical procurement criteria.
The broader direction is clear. Software quality engineering is moving from a gate at the end of development toward an intelligence layer that operates throughout the lifecycle.
Tricentis is betting that the next generation of testing will not simply automate existing test cases. It will use AI agents to explore software, evaluate other AI agents and help humans decide whether increasingly complex systems are safe to ship.
Market Landscape
The quality engineering market is being reshaped by two simultaneous forces: AI-generated software is increasing development velocity, while AI-powered applications are making traditional deterministic testing less sufficient.
Gartner’s 2026 research describes AI application and agent testing as an emerging discipline that enterprises still need to scale.
At the same time, McKinsey reports that most organizations remain early in enterprise AI scaling, despite widespread experimentation. Sixty-two percent of respondents said their organizations were experimenting with AI agents, while only about one-third reported that their companies had begun scaling AI programs across the enterprise.
That creates an opening for platforms that combine AI testing, agent evaluation, quality engineering, governance and release intelligence.
Tricentis is competing in a crowded ecosystem that includes traditional test automation vendors, developer platforms and emerging AI-agent evaluation companies. Its strategic advantage will depend on how deeply these capabilities integrate with enterprise application portfolios and whether customers can demonstrate measurable reductions in quality risk and release friction.
Top Insights
- Tricentis Aida autonomously explores applications, while AgentScore evaluates AI agents probabilistically, addressing new quality challenges created by agentic software.
- Release Risk Intelligence adds AI-driven risk prioritization to release management, helping enterprise engineering teams focus testing resources on the changes most likely to matter.
- Tricentis Labs gives customers early access to emerging AI technologies, creating a feedback loop between experimental capabilities and enterprise-ready quality engineering products.
- The company’s Tabnine acquisition expands its AI strategy beyond testing toward a broader software lifecycle spanning development, validation, governance and release.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
