Automation Centre turns sources into a structured dataset you can download. It is built for repeatable collection: the same sources, the same shape, a fresh dataset each time. ## A run in four steps 1. **Sources** — add web pages, or upload CSV, PDF, Word or Excel files. 2. **Plan** — describe the data you need. Review the extracted fields and the sources before approving. 3. **Run** — the approved plan is carried out and the records are produced. 4. **Download** — export the dataset and keep the run in your job history. ## Run limits One run is capped at **5 sources**, **12 pages** and **2,000 records**. The caps keep a single run predictable in time and cost. If you need more, split the work into separate runs. File uploads are limited to 10 MB per file and 25 MB combined for a job. ## Comparing runs Every run is saved. Open two runs to see what was added, removed or changed between them. This is the point of scheduling: the comparison shows the change, not just the current state. ## Repeating a run Choose **Repeat** on a completed job to schedule it again. A scheduled run appears in the job list with the date it is due, and the last run is linked so you can tell whether the schedule is keeping up. Scheduled dispatch depends on the site's scheduler running. If a scheduled job has not started, the job list shows it as due rather than pretending it completed. ## Exports Excel, CSV, JSON, text, Word and PDF. Word exports include the first 400 records and PDF exports the first 250, because of document size; use CSV or Excel for a complete dataset. ## What it is not It does not log in to websites, bypass access controls, or read anything you could not open yourself. It fetches public pages and files you provide.