Hermes trading agent setup with Hostinger and Telegram

This tutorial presents Hermes as a self-hosted trading assistant that users can message through Telegram. The speaker describes an open source framework with persistent memory, scheduled tasks and a loop that creates reusable skills. Claims that it improves over time refer to remembering context and building skills; the tutorial does not establish better trading returns.

Examples include a portfolio morning briefing, research into Nokia insider filings and a scan for potential long positions. A follow-up request produces pre-trade checks and a suggested trade ticket. The speaker notes that the Sunday example uses Friday's market information. Users supply their holdings, strategies and risk preferences before scheduling recurring reports.

The setup walkthrough deploys the Hermes template on a Hostinger VPS, opens its terminal and selects quick setup. It connects OpenAI Codex through browser authorization and selects GPT 5.5; the speaker says this model reaches usage limits sooner than GPT 5.4. Telegram configuration requires a bot created through BotFather, its token and the user's Telegram ID, followed by a connection test.

A dedicated local computer is another hosting option, but the demonstrated configuration uses a VPS and an external model provider. The speaker recommends a VPS for continuous availability and presents the trading examples as educational, without guaranteeing profits.