Set up a local agent for price watches, calls, and drafts

A poster keeps Hermes running locally to track price drops, make booking calls, and prepare emails for manual approval.

Done with Hermes Agent· Advanced· Nov 2026

What they did
They run Hermes on a 64 GB Mac Studio all day and night as a single entry point for scheduled tasks, with different specialist profiles behind it. For investing, they log each stock’s thesis, fair value, entry price, and trigger points, then have Hermes review prices and news on a timetable and explain any drop. For calls, Hermes passes restaurant reservations to Twilio with Vapi or ElevenLabs, waits for the other side to speak first, makes the booking, and reports back. For email, it writes messages through Microsoft Graph into Draft Cards, where they can edit them before sending.
How it went
It now takes care of repeat jobs such as price checks, booking follow-ups, and email drafts that wait for approval, without sending anything on its own. They say it has earned back its cost through improved decisions, while still stopping short of trading and asking before anything that costs money.
Tips
Keep a clear checklist for each stock or task so the agent knows what counts as a real trigger. Treat payments, calls, and sent mail as approval-only steps, and use cheaper models for simpler jobs if you are tracking monthly spend.

Try it yourself with Hermes Agent

You are my local agent running continuously on [computer/device]. Watch [stock, product, or website] for price changes on a schedule and tell me when the price falls below [threshold]. For each alert, explain what changed and why it may have happened based on recent news. Also prepare booking calls for [restaurant/business] using [Twilio/Vapi/ElevenLabs], but do not speak first, do not confirm any booking, and pause for my approval before any charge, payment, or final commitment. Finally, draft emails about [topic or sender] into [mail app/draft system] and stop at the draft stage; never send anything until I review the text and explicitly say to send.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on reddit.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.