pyne

Streamlining invoice processing saving 25 days per year

Streamlining invoice processing saving 25 days per year

Client Background

A confidential housing company processes a high volume of supplier invoices each month in e-conomic. Booking these to the correct accounts was a manual task taking 2-3 days for a finance controller.

Aspiration

Automate PDF invoice processing to cut manual effort while keeping human oversight and auditability.

How We Did It

  • Smart extraction with Python and LLMs: Parsed PDFs to capture supplier, lines, VAT, PO, and account hints.
  • Business rules and modeling: Normalized vendors, mapped GL logic, and validated totals to reduce exceptions.
  • e-conomic integration with approvals: Wrote entries to the ERP with an optional review step before posting.
  • Validation and scenarios: Reconciled to financials, added what-if views for pricing, mix, and cost changes.

Rollout framework:

  1. Strategy and alignment workshops with business and tech
  2. Technical discovery with domain teams
  3. Foundation rollout for infra, security, CI/CD, and branching
  4. Enablement and onboarding for users and developers
  5. Continuous improvement for migration, maintenance, and support

Migration at scale: An agentic migration assistant to port MS SQL objects into dbt, first 1:1, then automated refactoring, delivering an estimated 80% speed-up versus prior manual approaches.

Key Outcomes

  • 95% reduction in monthly time spent on invoice booking.
  • Fewer errors and rework due to standardized extraction and validations.
  • Clear audit trail with human-in-the-loop approvals.
  • Future-ready: A strong base for analytics, ML, and automation across multiple domains.
← Back to Cases