Agentic software development is reshaping how code is created, reviewed, and deployed, pushing the need for a more responsive testing framework. Traditional testing struggles to keep pace, relying on static test suites and ongoing maintenance, which can lead to false positives and missed bugs. Just-in-Time Tests (JiTTests) offer a solution. These tests are generated in real-time by large language models (LLMs) as new code is introduced, targeting regressions and unexpected behaviors without the burden of manual test authoring or maintenance. Catching JiTTests focus specifically on detecting issues linked to recent code changes. They use sophisticated methods to assess potential faults by simulating various scenarios based on the inferred intent of the code change. The process begins when new code is submitted, prompting the system to create "mutants"—versions of the code with intentional flaws. Tests are then generated and executed to identify any significant failures, allowing engineers to receive immediate feedback on relevant issues. This approach shifts the testing paradigm from a generic focus on code quality to a more targeted examination of specific changes. JiTTests eliminate the need for continuous test updates, as they are tailored to individual code changes and automatically adapt as the code evolves. Human oversight is only necessary when actual bugs are detected, dramatically reducing the time engineers spend on test maintenance and enabling them to focus on resolving real issues. This evolution in testing infrastructure aligns with the rapid pace of software development today, ensuring that testing keeps up with the demands of agentic coding.