AP Invoice Automation Hub for GCC Shared Services
Sapnity implemented an AI-enabled AP Invoice Automation Hub on Power Platform that sits above SAP, Oracle and Dynamics 365 finance cores — standardizing intake, approvals and exception handling for a 600+ FTE GCC shared services organization.
1. Business Problem — Email-Driven AP at Scale
The client ran a 600+ FTE shared services center processing invoices for multiple business units across North America and Europe. Finance cores were modern (SAP, Oracle, D365), but the “front door” of AP was still email, Excel and manual routing.
- Invoices arrived via 40+ mailboxes, supplier portals and scanned PDFs.
- Different BUs had different approval rules, tax treatments and GL coding patterns.
- 3-way match exceptions bounced between AP and local finance with no visibility.
- Vendors chased status by emailing individuals, not a structured channel.
- GCC leadership lacked a real-time view of AP load, cycle time and bottlenecks.
In practice, the GCC behaved like five different AP shops sharing one floor— each with its own macros and workarounds.
2. Sapnity’s Mandate
Leadership asked Sapnity to design an AP hub that would:
- Standardize AP intake and approvals without forcing ERP replacement.
- Embed AI extraction and rules in one place instead of per-BU macros.
- Provide a single workspace for AP agents across SAP, Oracle and D365.
- Expose live KPIs: cycle time, backlog, exception reasons, vendor SLAs.
- Set up a pattern the GCC can reuse for other finance processes (T&E, disputes).
3. Before — Fragmented AP Across ERPs
A single invoice could touch five tools before it reached the ERP posting step. Analysts downloaded PDFs from email, keyed them into different ERPs, and used Excel trackers to coordinate approvals and 3-way match issues.
- No unified queue — work was pulled from mailboxes, not from a central backlog.
- Invoice status lived in people’s heads and color-coded spreadsheets.
- Every BU had its own unofficial “rules” baked into personal macros.
- Month-end close required heroics to find stuck invoices on time.
The GCC was doing the right things — but the process lived in email and Excel, not in a system that could scale.
4. After — Sapnity AP Invoice Automation Hub
Sapnity introduced an AP Invoice Automation Hub on Power Platform: a single, governed workspace where all invoices enter, flow through AI extraction and rules, and then fan out to SAP, Oracle or D365 for posting.
Business Units & Vendors
Suppliers, local finance, and BU approvers sending invoices & approvals.
AP Intake & Triage App
Single front-door in Power Apps for email, portal and upload channels.
Invoice Intelligence Engine
AI/OCR extraction, duplicate checks, vendor and PO enrichment.
AP Workflow Engine
Approval routing, 2/3-way match handling, SLA timers and escalations.
ERP Posting Connectors
Standardized posting into SAP, Oracle and D365 Finance cores.
AP Analytics & Control Tower
Live backlog, cycle time and exception heatmaps by BU, vendor and ERP.
The same pattern is now used as the blueprint for new countries and BUs, with local rules configured in Dataverse instead of hidden in spreadsheets.
5. Implementation Story
Phase 1 — Discovery & Baseline
- Shadowed AP agents across SAP, Oracle and D365 workstreams for 2 weeks.
- Mapped real “street-level” flows: intake → coding → approvals → posting.
- Quantified rework and hand-offs at BU, vendor and exception-type level.
Phase 2 — AP Hub Blueprint
- Defined a single AP object model covering PO, non-PO, credit notes and prepayments.
- Designed a central AP queue with work-type tags, SLAs and skill-based routing.
- Agreed a minimum common rule-set; BU specifics moved into configuration tables.
Phase 3 — Hub Build & Pilot
- Built the AP Intake & Triage app with integrated AI extraction and validation.
- Implemented approval workflows with flexible levels and delegation.
- Rolled out to one high-volume BU (SAP) to de-risk the pattern.
Phase 4 — Multi-ERP Scale-Out
- Added connectors and posting patterns for Oracle and D365 Finance.
- Aligned exception categories and root-cause codes across all ERPs.
- Introduced a vendor-facing status view to cut down on chase emails.
Phase 5 — Optimization & Governance
- Set up monthly AP performance reviews using the new control tower dashboards.
- Trained GCC process owners to adjust rules and thresholds without IT releases.
- Created a playbook to onboard new BUs or acquisitions in 8–10 weeks.
6. Technical Architecture — Layered View
7. Reusable AP Automation Pattern for Future BUs
What the client gained was not just one project — but a repeatable AP automation pattern. New BUs and countries are now onboarded by configuring rules and mappings, not by building new workflows from scratch.
8. Outcomes & KPIs
| KPI | Before | After Sapnity AP Hub |
|---|---|---|
| Average AP cycle time (receipt → ready-to-pay) | 18–25 days | 7–12 days (by BU, with clear targets) |
| Invoices needing manual rework > 2x | ~40% | < 15% after rules & AI tuning |
| Exceptions without root-cause tag | ~60% | < 10% with mandatory coding |
| Vendor status chase emails | 7,000+ / month | < 2,000 / month with status visibility |
| Time to onboard a new BU | 6–9 months of projects | 8–10 weeks using the AP hub pattern |
The AP hub is now the GCC’s finance front-door — serving as a model for T&E, disputes and other shared services processes.
9. Sapnity Differentiators
- GCC-first design: We optimised for multi-ERP, multi-BU, India-based shared services realities — not a single-country AP team.
- Pattern-first, not project-first: Delivered a reusable AP hub template that reduces future onboarding from months to weeks.
- AI used where it matters: AI/OCR is wrapped with strong rule tables and human-in-the-loop checks, not left uncontrolled.
- Governed data model: Dataverse provides clean separation of concerns between rules, data and workflows.
- DevOps-ready: Managed solutions and ALM pipelines allow controlled moves from Dev → Test → Prod with GCC change governance.
For this GCC, Sapnity did not replace SAP, Oracle or D365 — we orchestrated them through a modern AP brain that finally made shared services feel truly shared.