Why this happens
Many established ERPs were built before open APIs were common. The data is there, but the only way out is through screens and manual exports. Teams end up spending hours every week downloading files and rebuilding the same reports.
How the automation works
- Sign in automatically. A browser automation tool such as Playwright logs in with a dedicated account, exactly as a user would.
- Extract. The job opens the required reports or screens and downloads or reads the data.
- Clean. A data step (for example with pandas) fixes formats, removes duplicates and joins tables so the numbers are consistent.
- Load. The clean data is stored and served to dashboards.
- Schedule and monitor. The whole job runs unattended on a schedule, usually daily, and alerts someone if a step fails, so a broken refresh never goes unnoticed.
What we built
For a distribution business, Kryyvo built an unattended Playwright job that signs in to the ERP and extracts the data every day; pandas cleans it and a Flask app serves 13 dashboards with 70+ KPIs, refreshed automatically every day. See the Distribution Analytics case study.
FAQ
Is browser automation reliable enough?
Yes, when it is built with care: stable selectors, retries, clear error alerts and a check that the data looks complete before it replaces yesterday's numbers.
What if the ERP screens change?
The job needs a small update when the ERP's screens change. Good monitoring means you find out on the first failed run, not weeks later.
Is it allowed?
It uses a normal user account to read your own company's data. Check your ERP licence terms and use a dedicated account with read-only permissions where possible.