An article that asks for an update.
Spot a broken reference or an outdated technical detail and send me a proposed correction with its source.
Into Telegram. With a topic, a draft and a question: ready to publish? This is the kind of automation I care about — less routine between an interesting idea and a useful result.
A visual example of the workflow, not a live Telegram message.
For me, AI is part of the working process. I use an assistant, automate the tracking of interesting articles and work with content that arrives in Telegram asking to be published.
The point is to shorten the distance between noticing something useful and doing something with it. A new article can become a topic to explore, an outline to develop or a draft to review.
The most valuable part is my perspective: what I agree with, what I would test and what my own projects have taught me. That is what turns a prepared text into an authored publication.
This is the editorial logic behind the workflow. Telegram is the point where preparation meets my attention.
Track interesting articles instead of repeatedly checking everything by hand.
Ask what makes this topic relevant to my website and its readers.
Turn the idea into a structure and a draft that can be reviewed.
Bring the prepared content into Telegram with a prompt to review it.
Check the material, add an authored perspective, then decide what to publish.
I want a prepared next step.
Not another tab to remember.
Automation can prepare the material. Editorial judgment gives it a reason to exist.
The topic should answer a real question or offer a useful perspective, rather than repeat a headline.
Check dates, claims and references. An outside article is a starting point for research, not a text to pass off as your own.
A draft can be corrected, postponed or rejected. A notification is an invitation to review, not proof that a text is ready.
These are ideas for extending the workflow, not claims about features already running on my website.
Spot a broken reference or an outdated technical detail and send me a proposed correction with its source.
With appropriate privacy rules, group recurring questions into topics worth explaining in a new article.
Compare controlled performance checks after updates and flag meaningful regressions with the affected page.
Prepare a short Telegram announcement, an email introduction and a social post for editorial review.
Suggest relevant internal links for a new publication and explain why each connection helps the reader.
Collect promising topics, unfinished drafts and items needing attention into one concise editorial briefing.
When designing this kind of automation, I start with the signal: what deserves a notification, what can wait for a digest and what should be ignored. A hundred alerts are just another kind of manual work.
The implementation also needs rules for duplicate materials, failed requests, access permissions and the use of external AI services. Publishing permissions and the approval step should be explicit.
How I connect services ↗What do you repeatedly read, copy, check or prepare?
Which websites, documents or services contain the information?
A digest, a draft, a CRM entry or a notification — what actually helps?
Decide what can happen automatically and what needs your approval.
An interesting topic, a content draft or a useful alert? Describe one repetitive task. We can turn it into a workflow with a clear result and a sensible place for your decision.
Discuss my automation ↗KATYUSHYN · HUMAN IDEAS, AUTOMATED ROUTINE