The always-on assistant | Master AI Automation in 4 hours Master AI Automation in 4 hours Course About Ayush Modules Sample chapter Toolbox The Microcap Minute Classroom / Module 06: Webhooks & Automation / Chapter 5 The always-on assistant Watch first, then read. Same lesson, your pace. What you will learn – Cron and scheduled workflows: AI that runs while you sleep – The daily digest pattern – Four guardrails that keep automations from embarrassing you From reactive to scheduled Webhooks react to events. The other half of automation initiates : runs every morning at 7, every Friday at 5, every hour. In n8n: swap the Webhook trigger for a Schedule/Cron node , same visual wiring. In raw terminals: cron , the Unix scheduler built into every Mac/Linux: crontab -e 0 7 * * * python3 /Users/ayush/api-lab/digest.py >> ~/digest.log 2>&1 Reads as: minute 0, hour 7, every day, run digest.py, append output to a log. Five fields (minute, hour, day, month, weekday) schedule anything forever, free, off-peak by design (pair with Module 4’s rule). The daily digest pattern The highest value-per-line automation most people ever build: Gather , pull yesterday’s items (RSS feeds, GitHub notifications, emails via filters, saved articles) Think once , send the pile to a free model (off-peak!): “Rank by relevance to my exam prep; summarise top 3 in two lines each” Deliver small , one Telegram/WhatsApp/email message: three bullets, links attached Ten inputs collapse into one 30-second read. The pattern generalises endlessly: weekly class-summary digests, price-watch alerts, revision-question-of-the-day. Guardrails: trusting machines wisely Always-on means unwatched, so build fences: Dry-run first , every new automation logs to a file for a week before it’s allowed to message humans Rate-limit yourself , max N messages/day; a broken loop must not text you 400 times Kill switch , one command/file/flag that stops everything ( touch STOP checked by the script) Log everything , every run appends to a log file; silent failures are the only unacceptable kind With those four, an overnight failure costs nothing but a log line. Without them, it costs reputation. Try it yourself Build the digest: Schedule trigger in n8n (or crontab + digest.py) pulling from two RSS feeds of your choice, ranked+summarised by a free OpenRouter model, delivered as one message. Run it in dry-run mode (log-only) today, then flip delivery on tomorrow morning. Add all four guardrails. Save the workflow export into learn/digest-workflow.json . Key takeaways – Scheduled triggers (n8n Schedule/cron) make AI initiate instead of react. – Gather → think once → deliver small is the digest pattern behind most useful bots. – Off-peak scheduling compounds free quotas (Module 4 meets Module 6). – Dry-runs, rate limits, kill switches and logs are what make unattended trustworthy. Download the exercise sheet (PDF) Module workbook (PDF) ← Prev: Roll your own receiver Module 06 index → Classroom / Module 06: Webhooks & Automation / Chapter 5 The always-on assistant What you will learn – Cron and scheduled workflows: AI that runs while you sleep – The daily digest pattern – Four guardrails that keep automations from embarrassing you From reactive to scheduled Webhooks react to events. The other half of automation initiates : runs every morning at 7, every Friday at 5, every hour. In n8n: swap the Webhook trigger for a Schedule/Cron node , same visual wiring. In raw terminals: cron , the Unix scheduler built into every Mac/Linux: crontab -e 0 7 * * * python3 /Users/ayush/api-lab/digest.py >> ~/digest.log 2>&1 Reads as: minute 0, hour 7, every day, run digest.py, append output to a log. Five fields (minute, hour, day, month, weekday) schedule anything forever, free, off-peak by design (pair with Module 4’s rule). The daily digest pattern The highest value-per-line automation most people ever build: Gather , pull yesterday’s items (RSS feeds, GitHub notifications, emails via filters, saved articles) Th
The always-on assistant
Written by
in