A ready-made system of 8 n8n workflows: collects RSS, summarizes via LLM, generates posts with emoji and images, publishes to Telegram on a schedule, and collects statistics in Google Sheets

You run a Telegram channel or help clients with content, but every day the same thing repeats: open a dozen RSS feeds, pick something interesting, rewrite it, come up with emoji, find an image, publish it, and then manually compile the stats. It eats up hours, and the result is inconsistent: sometimes a post gets forgotten, sometimes the image doesn't fit, sometimes the data gets lost.

The problem isn't that you work poorly. The problem is that the process isn't automated. And it's solved not by hiring another person, but by building a content factory on n8n. Below is a practical guide on how to build a system of 8 workflows that does everything on its own.

Why n8n and 8 workflows specifically

n8n is an open-source automation tool that supports complex scenarios, working with APIs, LLMs, and databases. It doesn't require programming, but it offers flexibility that simple "builders" don't have. Eight workflows isn't magic, but a proven decomposition: each stage of the content pipeline is placed in a separate scenario that can be triggered on a schedule or by an event.

This architecture provides transparency: if something breaks, you know exactly where to look. And it's easy to scale — add a new source, change the prompt, or connect a second social network.

Step 1. RSS collection: where to get content

The first workflow is the aggregator. It runs on a schedule (for example, every 30 minutes) and polls a list of RSS feeds: industry media, competitor blogs, topical digests. n8n works with RSS out of the box, and an additional Filter node cuts out duplicates and irrelevant entries by keywords.

It's important to set up deduplication right away: save the IDs or links of already processed articles in Google Sheets or a database. Otherwise, the next day you'll get the same posts.

Step 2. Summarization via LLM

The second workflow takes new articles and sends them to an LLM (for example, via the OpenAI API or another model). The task isn't just to retell, but to highlight the essence: 3–5 key ideas, facts, conclusion. A good prompt contains a role ("you are a Telegram channel editor"), length constraints, and a requirement to maintain a neutral tone.

The output is structured text that's easy to turn into a post. If the article is in a foreign language, the LLM will translate it right away.

Step 3. Generating posts with emoji and images

The third workflow is the creative one. It takes the summary and forms a ready post: headline, body, call to action, emoji for emphasis. The image is also selected here: either via a stock image API (Unsplash, Pexels) or via image generation from a description.

The key point is templating. You define the post structure once, and the LLM fills it in. This removes the risk of the model "making things up." Emoji are added by rules: no more than 3–5 per post, only for semantic blocks.

Step 4. Publishing to Telegram on a schedule

The fourth workflow is scheduled publishing. It receives ready posts and sends them to the Telegram channel via the Bot API. The schedule is configured flexibly: for example, 3 posts a day at 10:00, 14:00, and 19:00. n8n supports a queue: if a post fails to send, it will retry.

You can also add moderation here: before publishing, the post goes to a separate chat for approval. This saves you from accidental LLM errors.

If you don't want to build everything from scratch, there's a ready-made solution — Content Factory for Telegram: 8 n8n workflows turnkey. All eight scenarios are already implemented there, including publishing and stats collection.

Step 5. Collecting stats in Google Sheets

The fifth workflow is analytics. After publishing, it collects metrics: views, reactions, reposts, link clicks. The data is written to Google Sheets, where you can build pivot tables and charts. This provides feedback: which topics land, what time is best to publish, which sources get more engagement.

Automatic stats collection is what's most often ignored. And without it, the content factory works blind.

Step 6. The remaining three workflows: maintenance and scale

The sixth workflow is error monitoring. It watches the execution of all scenarios and sends a notification to Telegram or email if something fails.

The seventh is source management. Via Google Sheets you add or disable RSS feeds without editing the workflow.

The eighth is archiving. All published posts and metrics are saved to a separate table so you can analyze history and not lose data when clearing operational sheets.

What this looks like in practice

You wake up, and a post with an image and emoji has already been published to the channel. A row with stats has appeared in Google Sheets. You don't waste time on routine, but focus on strategy: choosing new sources, testing formats, negotiating ads.

For micro-agencies, such a content factory can be put on a production line for clients. And if you work with short videos, check out the short video pipeline on n8n — it automates voiceover, editing, and publishing to TikTok, Reels, and Shorts.

What's important to consider

Don't try to automate everything at once. Start with one channel and three sources. Make sure the prompts produce consistent results and publishing doesn't fail. Then add new feeds and platforms.

Keep an eye on quality. LLMs can make factual errors, so for sensitive topics, set up manual moderation. And don't forget about image copyright — use stock APIs or generation.

Conclusion

A system of 8 n8n workflows isn't just automation, but a change in approach: you stop being a "manual editor" and become an operator of a content factory. RSS, LLM, post generation, scheduled publishing, and stats in Google Sheets work together, saving dozens of hours a month.

You can view and grab a ready-made set of such workflows at the link. It's the fastest way to launch a content factory without building it yourself.