Personal Workflow automation on local device set up.

Set up a local, always-on AI “supercomputer” at home that keeps track of messages and notifications, manages scheduling, reviews and triages email, and produces a daily recap of new oncology data — all running on hardware you own, with no cloud dependency for sensitive data. Model runner: Ollama or NVIDIA's own TensorRT-LLM stack (pre-installed on DGX OS) to serve a local LLM (e.g., a 70B-class open model). • Interface: Open WebUI as the chat/dashboard front end for interacting with the assistant from any device on the home network. • Orchestration: n8n or a lightweight agent framework (e.g., LangGraph-based) to connect the model to email, calendar, and news-feed sources and run scheduled jobs. • Connectors: IMAP/Gmail/Outlook API for email, CalDAV/Google/Outlook API for calendar, and PubMed/ClinicalTrials.gov/ASCO Daily News feeds for oncology data. Need How the device handles it Messages & notifications Local assistant (e.g., Open WebUI + Ollama, or a small agent framework) polls connected accounts and surfaces a filtered, prioritized feed instead of raw notifications. Scheduling Calendar connector (CalDAV/Google/Outlook API) feeds a local agent that flags conflicts, preps daily agendas, and can propose meeting times — all processed on-device. Email review IMAP/API connector pulls new mail; local LLM triages by sender/urgency and produces a daily digest, similar to open-source tools such as the Nextcloud AI Assistant or custom LangGraph/n8n pipelines. Daily oncology data recap A scheduled job pulls new items from sources like PubMed, ClinicalTrials.gov, and ASCO Daily News, then the local model summarizes and ranks them into a short morning brief (the published “Daily Dose” research pipeline is a public example of this pattern). integrate clinical trial oncology NCT sites locally in new work as well as top five national cancer centers