5 Things I Automate with My AI Agent Every Day
I’ve been running a PolymathClaw instance for a while now. Here are five automations that actually stuck — things the agent does every day without me thinking about it.
1. Morning briefing at 7:30am
Every weekday at 7:30am, my agent:
- Checks the top stories on Hacker News
- Searches Google News for topics I care about
- Checks the weather for my city
- Pulls my Todoist tasks for the day
It compiles everything into a short briefing and sends it to me on Telegram. By the time I’m drinking coffee, I know what’s happening and what I need to do.
How I set it up: I just told my agent: “Every weekday at 7:30am, check Hacker News, Google News for AI and startups, the weather for Denver, and my Todoist tasks. Send me a short summary on Telegram.”
One message. Done.
2. Task management through chat
I stopped opening Todoist directly. Instead, I message my agent:
- “Add a task: review the proposal by Thursday”
- “What’s due today?”
- “Move the report task to next Monday”
- “Done with the invoice”
It handles the Todoist API calls. I just talk to it like I’d talk to an assistant.
The nice part: it works from Telegram, so I can add tasks from my phone without opening another app.
3. Research on demand
When I need to look something up, I ask my agent instead of opening a browser:
- “What are the current pricing tiers for Vercel?”
- “Find me three recent articles about AI regulation in the EU”
- “Compare the specs of the M4 MacBook Pro vs the M3”
It opens its browser, visits the relevant sites, and gives me a concise answer. No ads, no cookie banners, no scrolling through SEO-optimized filler content.
For deeper research, I ask it to compile a report: “Research the top 5 project management tools for small teams. Compare pricing, features, and user reviews. Save the report to my files.”
The report shows up in my dashboard’s Files tab, ready to download.
4. Email drafting
I don’t use the Gmail skill for reading emails (I prefer to check email on my own schedule). But I do use it for drafting:
- “Draft a reply to John’s email about the project timeline. Tell him we’re on track for the March deadline and ask if he wants a progress call next week.”
The agent writes the draft and I review it before sending. It knows my writing style from previous conversations, so the drafts sound like me.
5. Competitor monitoring
Every Monday, my agent checks a few competitor websites and tells me if anything changed — new blog posts, pricing changes, new features. It’s a 30-second Telegram message that would take me 20 minutes to do manually.
Setup: “Every Monday at 9am, visit [competitor URLs]. Check for new blog posts, pricing changes, or product announcements since last week. Send me a summary.”
Why these work
These aren’t ambitious AI workflows. They’re small, practical automations that save 30-60 minutes a day combined. The key is that they run automatically — I don’t have to remember to check anything.
The agent lives in Telegram, so I interact with it the same way I message a friend. No app to open, no dashboard to log into, no workflow builder to configure.
If you want to try setting up similar automations, start with a free trial. The first 30 messages are free, no API key needed.
Try PolymathClaw free
30 free messages, no credit card. Full plan: $19/mo with 7-day trial.