Dunning ensures invoices are paid on time, cash flow stays healthy, and customer relationships remain intact. Accuracy, consistency, and transparency are essential.
Here at KNIME, we redesigned the dunning process, adding in AI, to be robust, scalable, and auditable, without losing the personal, individualized communication KNIME customers respond to.

The dunning process challenge
- The dunning process was manual, fragile, and time-consuming.
- The team relied on multiple Excel sheets and Word documents, searched through inboxes to reconstruct email histories, and manually tracked overdue invoices.
- There was a lack of transparency, which made the process hard to scale and difficult to audit.
The goal: A robust, auditable, and scalable dunning process
Create a robust, auditable, and scalable process by automating the dunning process without losing the personal, KNIME-style communication customers respond to.
We team wanted to:
- Eliminate manual Excel and Word files
- Centralize invoice and email history
- Use AI to support, not replace, human decision-making
The solution: Manual effort down from ~15 to ~3 minutes per case
With KNIME, we redesigned the entire dunning process end-to-end:
- Data from Odoo (invoices, customers) and Zendesk (email history) is automatically collected and integrated into the solution
- Overdue invoices are monitored continuously
- Weekly Slack alerts notify accountants of status changes
- Shareable management KPIs and AR reports via data apps
- Operational views of overdue invoices with full email history and AI generates personalized email suggestions, tailored to each customer and conversation context
The impact
- Scalable and auditable dunning process
- Significant reduction in manual work
- More consistent, personal customer communication
- Earlier detection of overdue invoices
- Reliable, automated reporting for finance and management
The dunning process before vs after KNIME
| Before | After |
5 manually maintained Excel sheets | Fully automated Accounts Receivable reports |
Manual email searches | Centralized email history |
High effort per invoice ( ~15 min) | ~3 minutes per case |
Inconsistent monitoring | Automated Slack alerts |
Hard to scale or audit | Transparent, documented process |
Key takeaway
This project showed that successful AI adoption isn’t about replacing people, but about redesigning processes, empowering teams, and using AI where it truly adds value.
