Every day our system repeats the same routine: new information arrives through our API, records must be updated, customers require quick responses, and performance dashboards need to be checked. I want a single AI-driven agent that handles these three strands—data entry, customer support, and system-performance monitoring—without human intervention. Here’s the flow I have in mind. The agent polls the API on schedule, validates the payload, and inserts or amends the relevant records. While it works, it should watch for anomalies in CPU, memory, or response times, flagging anything abnormal in a Slack or email alert. On the customer side, the agent will read incoming tickets, suggest or send context-aware replies, and escalate anything outside predefined confidence levels. At the close of each cycle it will compile a concise report: updated record counts, uptime metrics, customer-support KPIs, and a short trend analysis showing spikes or deviations. A Python stack with REST integration, async handling, and libraries such as pandas, LangChain or similar for LLM prompts would suit us, but I’m open to other reliable options if they meet the same goals. Execution can be triggered by cron, a lightweight scheduler, or a serverless function; what matters is that it runs unattended and keeps detailed logs. Deliverables • Production-ready agent script(s) with clear setup instructions • API interface module with error handling and retries • Automatic record-update routine tied to the incoming data schema • Customer-support responder that uses an LLM and fallback rules • Monitoring hooks plus alert channel integration • Daily report generator (CSV and human-readable summary) • Read-me covering deployment, configuration, and how to extend rules Success is a system that I can install once and trust to repeat these actions every day, with the log and report showing me it all worked.