Hermes Agent setup: VPS, Telegram and scheduled tasks

This beginner tutorial sets up Hermes Agent on a Hostinger VPS and connects it to hosted models through Nous Portal. The presenter starts with DeepSeek V4 Flash, then explains model selection, reasoning effort and fallback chains. Although Hermes can run on a personal computer, this walkthrough uses a server that stays awake for scheduled work. It does not demonstrate local model inference.

The setup connects Telegram and the Hermes desktop app to the same server. The desktop demonstration uses macOS, with Windows shortcuts also explained. Conversations remain separate threads with shared memory, and a handoff command moves a session between interfaces. The presenter notes that Telegram bot chats lack end-to-end encryption.

Google Workspace setup covers API activation and OAuth credentials. Examples include summarizing a PDF, exchanging documents through Drive, and filing Gmail receipts into Sheets. The presenter recommends a separate Google account and creates recurring jobs paused for review before activation.

Later sections explain saved preferences, reusable skills and SOUL.md personality settings. Temporary sub-agents handle parallel research, while persistent profiles take roles on a Kanban project board. The guide also covers usage costs and troubleshooting. The desktop app and server backend require separate updates.