Hermes connects research briefings and work logs
Qian reports using Hermes for research briefings, retrospective work summaries and local writing tasks delivered to his phone through Feishu/Lark.
guochengqian.github.io
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
At [morning time], research [my topics] and send a brief with primary links and relevance. At [evening time], inspect my work sessions and list progress, decisions and next actions. Prepare writing as drafts for my review.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Gordon Guocheng Qian
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Qian reports using Hermes for research briefings, retrospective work summaries and local writing tasks delivered to his phone through Feishu/Lark.
- Human involvement
- Qian supplies research interests and writing intent, and manually reviews blog drafts before posting. His grammar-polishing instructions preserve technical meaning.
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
- Run Hermes beside local research repositories on a Mac mini.
- Schedule a morning briefing with research interests and direct paper links.
- Use session search for nightly summaries and weekly themes.
Gordon Guocheng Qian runs Hermes on a Mac mini alongside his paper repositories, notes and blog files. His recurring workflow combines an outward-looking research briefing at 8am with work summaries at 10pm. Feishu/Lark carries the results to his phone, so the same assistant remains reachable away from the computer.
The morning job searches for papers, reports and model releases relevant to his interests, including video generation and multimodal systems. He asks for direct paper links and an explanation of why an item matters to his research. Evening jobs use session search and memory to reconstruct manuscript edits, experiments, decisions and unfinished work. A weekly summary groups those activities into broader themes.
Qian also gives the agent access to local LaTeX files for polishing and consistency work. For blog posts, he begins with rough notes, asks for a structured draft, then manually reviews the final HTML or Markdown before posting. His reported benefit is continuity across fragmented research tasks, rather than a measured productivity gain. The useful pattern is to specify interests, report format and carry-over questions in the scheduled prompts, while retaining review of technical claims and publication.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · Gordon Guocheng Qian
Frequently asked questions
What makes the briefing specific to his research?
The prompt names research interests and asks why each paper or release matters, with direct source links rather than a general news list.
How does Hermes reconstruct the day?
The evening jobs use session search and memory-aware summaries of actual edits, experiments and decisions.
Who approves publication?
Qian manually reviews final blog drafts before posting; the account does not establish unattended publication.
More details
Remaining steps
- Review prepared writing before publishing it.
About this account
Scheduled jobs select relevant research, inspect session history and prepare reports; file and terminal tools support local writing tasks between user messages.
Firsthand account by Gordon Guocheng Qian; original publication date verified against Primary page publication date. Outcomes are the author’s reports, not independently reproduced. Suggested prompt written by Agents In Use; not the author’s original prompt.
Credit: Gordon Guocheng Qian
Source published
Added to Agents In Use
Source checked
Tools and topics
Hermes cron · Mac mini · Feishu/Lark · LaTeX
Research briefing · Work logs · Local files
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.