Why I started using it
I was spending too much time extracting data, arranging it in Excel, and cleaning it. I started looking for a better way to handle those repeated steps. That led me towards AI, coding, and n8n.
What I built
My local research-to-content prototype connected source collection, structured records in Notion, and Gemini-assisted drafts. n8n coordinated the steps. The saved workflow documents that setup; I have not retested it as a current production service. In 2025, I also built an error-triggered workflow that asks Claude for an analysis and saves a cause analysis and suggested fix prompt in Notion.
How I would keep it under control
I would keep the source link, publication date, extracted facts, and draft in separate fields. I would check for duplicate records, missing data, and failed steps. A draft should wait for my review before it is published.
Where I want to take it next
I want to use scheduled follow-up to compare model assumptions with new results and collect related company news. I would check that periods, units, and definitions match before interpreting a difference. The workflow can gather the information; I still need to explain what it means.
How I would measure whether it helps
I would run the same task manually and through the workflow, then record total time, correction time, missing records, and duplicate records. I have not yet measured a reliable time saving. A faster run only helps if the result is also usable.