AI Layer

AI in Converged is not a separate chat added for appearance. It is a control layer over data, interface, and processes. A user can ask what is happening with an order, request a client reply, find a delay, launch an allowed workflow, or collect a summary of machine load.

The model receives context from the platform: requests, statuses, files, telemetry, client history, access rights, and current tasks. The answer is therefore based on company data, not generic reasoning. If an action is needed, AI does not bypass the system directly: it calls allowed functions, microservices, or workflows through a controlled layer.

Model providers connect through adapters. A single installation can use GPT, Claude, DeepSeek, Mistral, Gemini, or other engines if they fit the task and the client’s policy. One model can communicate with clients, another can parse technical requirements, and a third can analyze documents or production statuses.

The key principle is control. AI has its own access profile, every action is logged, and critical operations run under the same rights and policies as human actions. This allows agents to handle routine work without giving them uncontrolled power over production.