1.19.0 — AI-Driven Test Management

AI-Driven Test Management. A new chapter brings the AI capabilities together and explains the idea behind them: a stable, deterministic execution core with an agentic layer on top, and an autonomy spectrum you choose where to sit on. See AI-Driven Test Management.

AI Assistant. The in-app, agent-style chat that creates, fixes and reasons about a test over a persistent, resumable session — with per-change approval, auto-approve, cost tracking and a tenant-wide session overview. See The AI Assistant.

AI-driven onboarding and the Application Map. Point an agent at an application and it scans the screens into a visual Application Map, proposes test coverage for you to review, and builds the tests you accept. The thinking behind it is in The idea behind onboarding flows.

Knowledge Base. A per-application library of markdown articles that gives the AI lasting context; the index article is injected into every assistant session automatically, and a scan seeds it for you. See Knowledge Base.

MCP server (developer integration). Connect Claude Code, Claude Desktop, Cursor, Windsurf or VS Code Copilot straight to your tenant from your editor and drive the platform via the Model Context Protocol. See MCP (Model Context Protocol).

Applications chapter. "Application Settings" is now its own Applications chapter, restructured around the application detail's four tabs (General, Knowledge Base, Application Map, Settings).