genumis
All articles
4 mingenumis team

Case Study — Vitawell.tech: Daily News Analysis and Blog Content with AI Agents

The problem we solved

The Vitawell.tech blog team needs to track new publications and research from leading outlets every day and turn them into short digests covering the areas relevant to the blog. Done manually, that means a daily sweep through a list of sources, fact-checking, and rewriting — routine work we handed over to an AI agent.

The AI agent's daily process in Claude Cowork

Every day, on a schedule, an AI agent (Claude Cowork) starts up and runs the daily-research skill from a custom plugin. Here is what the process looks like, step by step.

Step 1. Checking existing articles in Airtable

First, the agent checks Airtable, where all previous articles and intermediate results are stored, to make sure it doesn't pick up a topic that has already been covered.

The agent checking existing articles in Airtable

Step 2. Scanning sources for new publications

The agent goes through the list of sites defined in the skill, looking at the previous day, and checks whether any new articles fit the blog's categories. The results appear right in Claude:

Source scan results shown in Claude

For every match, the agent shows a short summary and waits for confirmation: the user picks which article to work on today. This is the only point in the process where a human is needed.

Step 3. Research and keywords

This time we picked article #4 from the research. The agent moves on to studying additional materials, and the Progress panel shows its plan for what comes next: draft the structure, write the article (saving everything to Airtable in parallel), and trigger the n8n workflow.

Researching additional materials, with the plan visible in the Progress panel

Step 4. Article structure

Based on the research results, the agent drafts the structure of the future article:

The drafted article structure

Step 5. Publishing via n8n → Ghost CMS

Once the text is ready, the n8n workflow takes over: it calls an AI agent that writes a cover-image generation prompt based on the finished text and saves it, after which the article is published to Ghost CMS.

The n8n workflow that publishes the article to Ghost CMS

Step 6. Social media copy and Buffer scheduling

After publishing, the agent prepares a short social media text (which is also saved), and then schedules the posts via Buffer MCP:

Scheduling social media posts via Buffer

Tools used in the AI workflow

Diagram of the tools in the AI workflow

What human work was eliminated

The human kept exactly one decision in this process — which topic to take on. Everything else, from research to scheduling social posts, happens automatically.