Commit

AI applications are becoming more than a prompt and a response. The interesting applications increasingly need to make decisions, use tools, evaluate their own work, branch into different paths, wait for people, recover from failure, and continue where they left off. Laravel already provides the application infrastructure to run those systems. What it has lacked is a native way to model the AI work itself as an execution graph. LangGraph demonstrated this model in the Python ecosystem: AI work expressed as graphs of nodes, edges, state, routing, cycles, and interrupts. AgenticGraph brings that class of orchestration into Laravel and PHP, so increasingly sophisticated AI workflows do not have to become a separate Python service simply because they outgrow a linear request. Coming this week