TestMu AI, formerly LambdaTest, is expanding its agentic software testing platform with an Assurance Lifecycle for Kane CLI, shifting the tool from AI-assisted browser test creation toward requirement-driven quality engineering. The new workflow turns product specifications into traceable acceptance criteria and runnable tests, records execution evidence, measures proven versus missing coverage, and flags tests as stale when underlying requirements change.
AI coding agents are making software development faster, but they are also creating a new problem for engineering teams: keeping validation synchronized with code that can change faster than conventional testing processes.
TestMu AI is targeting that gap with the Assurance Lifecycle, a new set of commands in Kane CLI designed to connect software requirements directly to test design, execution and evidence.
The company, formerly known as LambdaTest, describes the release as an expansion from browser automation into what it calls agentic quality engineering. Instead of simply generating tests from a natural-language prompt, Kane CLI can now ingest requirement documents, derive proposed use cases, turn approved requirements into acceptance criteria and runnable tests, and maintain the resulting test suite as specifications change.
The new capabilities are available in Kane CLI 0.6.1 and later. (testmuai.com)
That distinction addresses a longstanding problem in software quality: a test suite can show that a collection of tests passed without demonstrating which product requirements those tests actually verified.
From test generation to test assurance
Kane CLI launched in April 2026 as a terminal-native browser automation tool designed for both developers and AI coding agents. It uses natural-language descriptions rather than requiring users to construct browser selectors manually and can run tests locally, headlessly and in CI environments. (testmuai.com)
TestMu AI subsequently added kane-cli generate, which creates test scenarios from a plain-language description.
The Assurance Lifecycle takes a different starting point: the actual requirement.
A product team can use kane-cli context ingest to snapshot a PRD, specification or policy document into a local content-addressed store. An agent then extracts proposed use cases and cites the specific source material it used.
Those proposals remain untrusted until a human reviews them.
Teams can promote, edit or reject the extracted use cases before moving to the design stage. That review mechanism is central to the architecture because it prevents an AI-generated interpretation of a requirement from silently becoming the basis for production tests.
Once approved, kane-cli design tests converts use cases into acceptance criteria, scenarios and runnable tests. TestMu AI says each scenario receives exactly one corresponding test, with every test tagged to the criteria it is intended to verify. (testmuai.com)
The result is a traceability chain connecting a source requirement to an executable validation artifact.
Measuring what was actually proven
The next part of the lifecycle addresses another weakness in conventional test reporting: the difference between the number of tests that exist and the requirements those tests actually prove.
Kane CLI’s cover command reports coverage across two separate dimensions.
The first is execution depth—what a sealed evidence pack actually proved during a test run. The second is design completeness—what the test design still needs to verify.
That means a passing test suite does not automatically translate into complete requirements coverage.
A product could have dozens of successful tests while still missing a critical acceptance criterion. Kane CLI records that missing requirement as a ranked coverage gap rather than treating the absence of a test as equivalent to coverage.
The evidence pack itself is sealed after execution, creating a record from which TestMu AI says coverage can be measured without recomputing the result later. (testmuai.com)
For organizations adopting AI-generated development workflows, that distinction could become increasingly important.
Gartner says enterprise AI coding agents are moving beyond code generation into broader software-development lifecycle activities, with planning, creation and review increasingly handled by agentic systems. The research firm predicts that by 2027 more than 65% of engineering teams using agentic coding will treat integrated development environments as optional, shifting more control, governance and validation toward automated platforms.
As the amount of AI-generated code increases, testing becomes less about producing more test cases and more about establishing whether those tests provide defensible evidence that software meets its requirements.
Maintaining tests as products change
The final component of the Assurance Lifecycle is maintenance.
When a requirement document changes, kane-cli maintain reconcile identifies the affected material and creates an update plan. Tests and use cases derived from an earlier source snapshot can be marked stale until they are re-verified. (testmuai.com)
That is intended to address test drift, where a suite continues to pass against software even though the requirements that originally justified those tests have changed.
The approach creates a feedback loop rather than a one-time generation process:
requirements → use cases → acceptance criteria → tests → evidence → coverage → maintenance.
Each stage is represented by a separate command, allowing engineering teams to stop for review before continuing downstream.
The assurance store remains in the project directory and uses an append-only, replayable history. TestMu AI says the sources, use cases, designs and review decisions remain local rather than being stored on a remote server.
The workflow can also run headlessly in continuous integration environments or through an AI coding agent. (testmuai.com)
Validation becomes part of agentic development
The timing reflects a larger shift in software engineering.
AI coding agents are increasingly capable of implementing features, modifying repositories and working through multistep development tasks. GitHub’s coding agent, for example, can work asynchronously on assigned issues, create pull requests and run tests before requesting human review.
The resulting development model increases the importance of the validation layer.
Gartner has warned that extensive use of AI coding agents can contribute to architectural technical debt and says engineering teams need controls to manage the quality and maintainability consequences. The research firm has also identified agentic testing as an emerging requirement because agent behavior and software workflows can become more complex and less deterministic.
TestMu AI is positioning Kane CLI against that problem by making the requirement itself part of the test system’s data model.
This is a meaningful distinction from generic AI test generation. Generating ten plausible test cases can accelerate a QA team’s work, but it does not necessarily answer whether those tests correspond to the actual product contract.
The Assurance Lifecycle is designed to make that relationship explicit.
Competing in an agentic quality market
The move also places TestMu AI in a rapidly developing category of AI-augmented software testing.
Gartner defines AI-augmented testing platforms as tools capable of automating or orchestrating activities including test generation, maintenance, optimization, prioritization, analysis and test-value scoring, increasingly alongside AI coding assistants and DevOps platforms.
That market includes established testing and DevOps vendors as well as newer agentic quality platforms.
TestMu AI’s differentiation is increasingly centered on the terminal and on AI-agent workflows. Kane CLI can interact with coding agents such as Claude Code, Cursor and other development environments, while the Assurance Lifecycle gives those agents a structured mechanism for deriving tests from requirements and reporting evidence.
The company is not positioning Assurance as a replacement for kane-cli generate. Instead, the two workflows serve different purposes: generation provides fast exploratory test ideas, while Assurance is designed for teams that need traceability and an auditable explanation of what has actually been verified.
That distinction could become more important as engineering organizations move toward agent-driven development.
If AI agents can produce software faster than teams can manually review every change, quality systems need to move from simply generating more tests to establishing a reliable chain between what the product is supposed to do and what the software has actually demonstrated.
TestMu AI’s latest Kane CLI release is built around that premise.
Market Landscape
AI coding is expanding beyond autocomplete into agentic development workflows, increasing pressure on software teams to automate validation alongside implementation. Gartner says the AI-augmented testing market is evolving toward autonomous capabilities covering test generation, maintenance, prioritization and analysis.
At the same time, Gartner warns that AI coding agents can introduce technical and architectural debt and recommends organizational controls and verification mechanisms before scaling SDLC automation.
TestMu AI’s Assurance Lifecycle targets this gap by linking requirements, test design, execution evidence and maintenance in a single local workflow.
Top Insights
- TestMu AI’s Assurance Lifecycle connects product requirements directly to acceptance criteria, runnable tests and execution evidence.
- Kane CLI separates AI-derived proposals from trusted artifacts, requiring human review before downstream test design proceeds.
- Coverage reporting distinguishes what an execution actually proved from requirements the current test design still lacks.
- Requirement changes can automatically make dependent test artifacts stale until engineering teams re-verify them.
- The release reflects a broader shift toward agentic software development in which automated validation becomes part of the SDLC itself.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
