Hermes prepares reviewed Facebook outreach
Drew Carr describes using Hermes to find relevant Facebook posts and draft useful replies, with human approval before posting as part of an email-list growth workflow.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Find authorized community posts where I can give a useful answer. Explain relevance, prepare a reply in my voice and show any proposed business link. Wait for my approval before posting.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Ryan Doser interviewing Drew Carr
- Original account
- Read the source ↗
- Published
- 8 Jun 2026
- Source link checked
- 15 Oct 2026
- Reported result
- Drew Carr describes using Hermes to find relevant Facebook posts and draft useful replies, with human approval before posting as part of an email-list growth workflow.
- Human involvement
- Carr reads and approves replies before posting and writes actual cold emails himself. Subscriber growth is reported without a causal experiment.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Carr directs Hermes through Telegram while a separate laptop runs browser automation.
- Hermes finds relevant Facebook-group posts and judges whether a helpful reply fits.
- It drafts a response using Carr's context and prior interactions.
Ryan Doser's account of an AI Rabbit Holes interview records Drew Carr's Hermes outreach setup. Carr controls the agent through Telegram while browser automation runs on a separate laptop. The reported business goal was growing the audience for ClawDrop.
One workflow searches relevant Facebook groups for posts where Carr can help. Hermes judges whether a post is relevant and whether mentioning the business would fit a useful answer. Plain scripts handle predictable browser operations, while the model supplies the judgment and reply draft. The account says Carr reads each proposed comment and approves it before it is posted.
The same interview describes a separate cold-email pipeline involving prospect lists, address verification, qualification and campaign storage. Carr still writes the actual emails. Those details matter because describing all the growth as agent-written email would misrepresent his role.
The account reports reaching a thousand subscribers in eighty days without advertising spend, but does not establish a controlled causal contribution from any one automation. Its stronger evidence is the concrete division of work: an agent gathers context and drafts, scripts operate the browser, and Carr checks whether each reply makes sense before external posting.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: ryandoser.com · Ryan Doser interviewing Drew Carr
Frequently asked questions
Who approved Facebook comments?
The account says Carr read and approved each draft before it was posted.
Did the agent write all his cold emails?
No. The interview says Carr still writes the actual email copy.
Can the subscriber growth be attributed to this agent alone?
The source reports growth but does not isolate the contribution of a particular workflow.
More details
Remaining steps
- Carr reviews and approves each comment before posting.
About this account
Carr directs Hermes through Telegram while a separate laptop runs browser automation. Hermes finds relevant Facebook-group posts and judges whether a helpful reply fits.
Firsthand account opened and read. Outcome is source-reported, not independently reproduced. Suggested prompt written by Agents In Use. Primary interview publisher account; subscriber totals are guest claims. No independent attribution of growth to automation.
Credit: Ryan Doser interviewing Drew Carr
Source published 8 Jun 2026
Source checked 15 Oct 2026
Tools and topics
Hermes Agent · Telegram · Browser automation
Hermes Agent · Telegram · Browser automation
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.