I need production-grade agentic AI components that can drop straight into my custom in-house platforms and run entire workflows without hand-holding. The agents must remember past interactions, draw on tool libraries, call external or internal APIs, and—most importantly—see a process through from the first trigger to final completion entirely on their own. Here is the baseline I expect them to hit: • Persistent memory so each agent builds context over time rather than starting cold on every call. • Direct tool use (file handling, data transformation, email dispatch, scheduling, etc.) triggered by its own reasoning loops. • Seamless API integration so the agent can query or update any service my stack exposes, as well as third-party endpoints. • Autonomous decision logic that lets it branch intelligently when unexpected states appear, yet always realign with the primary goal. • Full end-to-end workflow handling—this is the non-negotiable core of the project. I will provide a detailed specification for each agent—inputs, expected outcomes, edge cases, and success metrics—once I know you can architect, engineer, and ship these capabilities reliably in a production environment. If your expertise covers persistent-memory frameworks, reasoning-loop orchestration (e.g., LLM-based planners or symbolic hybrids), secure API gateway design, and robust fail-safe mechanisms, let’s talk through the next steps.