Automate a local briefing bot for news
Hermes Agent watches feeds on a small local server, filters stories, and posts short digests to Telegram on a schedule.
- What they did
- The author had Hermes Agent running on a Proxmox virtual machine with 4 vCPU, 8 GB of RAM, and no GPU, while Qwen 3.6 MoE was served locally from their RTX 5060 Ti. They started it with one Telegram instruction to make a daily 8 AM cron task, and the agent handled the scheduling, checked RSS and MCP sources, and matched new items against its memory before sending updates to Telegram topics.
- How it went
- The setup produces a daily briefing plus a startup and VC roundup at 9 AM, and it does so without API spend. After a few weeks, its memory began supporting trend-aware summaries, and Hermes also created reusable skills. The post did not include formal performance scores.
- Tips
- A simple goal worked better than a highly detailed set of instructions. The faster MoE model outpaced the dense 27B model, which was about 3x slower.
Try it yourself with Hermes Agent
Act as my always-on news curator. Use [RSS feed URLs] and [other source types] to track stories about [topics I follow]. At [time or interval], pick the [number] most useful items, remove repeats and low-value posts, and send a brief digest to [Telegram or other destination]. Before you create or change any schedule, send any message, delete anything, submit anything, or make any payment, stop and ask for my approval first. Then generate a test digest so I can check the format.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
2 more reports of this
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Automate a weekly competitor scan that keeps learning
A product manager had Hermes generate a recurring competitor briefing that got quicker and uncovered patterns he had missed.
Turn voice memos into structured Obsidian notes
Have an agent convert spoken Telegram messages into Obsidian notes with a transcript, a tighter version, linked context, and a draft.
Create a survey form and draft the questions
A developer used Manus to draft a Google Form for undergraduates’ opinions on AGI, then ran the script manually to create it.
Start desktop research work from your phone
They used Claude Cowork to begin research on a machine through a phone chat and get a summary back later.
Use Codex to summarize Logseq journals from the CLI
They used Codex with the Logseq CLI docs to query a note graph and turn months of journals and tagged pages into summaries.
Set up a daily Reddit and competitor scan
They use Cowork to gather relevant tech posts, market signals, and draft replies without checking everything by hand.