From manual accounts payable to AI-powered finance in 12 Weeks

Business challenges

Shell's finance teams were responsible for receiving invoices and supporting documents, entering information, validating transactions, managing approvals and preparing invoices for posting into SAP, this was manual work across the full invoice lifecycle.

Different business scenarios required different validations, documents and workflows, including:

• Non-PO invoices
• PO-based goods invoices
• Service invoices
• Service Entry Sheets
• Goods receipt and delivery documentation
• Tax validation
• Approval workflows
• Supplier master data
• Custom SAP processes

As implementation began, automation exposed weaknesses the manual process had been absorbing. Supplier records were missing information such as email addresses and NTN details, business rules were not standardised, and years of customisation inside SAP had added further process complexity.

Fixing this required Finance, IT and the business to simplify processes, improve data quality and agree on how the future operating model should work, alongside deploying the technology itself.

Why SpendConsole was chosen

Shell needed a solution capable of orchestrating the Accounts Payable process end-to-end, across the full range of invoice scenarios it ran through SAP.

SpendConsole provided a platform that combined AI-powered document processing, business rules, workflows and SAP integration, built for the realities of Shell's operating environment.

The approach centred on several principles:

• End-to-end automation, not invoice capture alone.
• AI-powered document processing across invoices and supporting documentation.
• Integration with SAP that preserved the organisation's core ERP environment.
• Business-rule automation for different invoice and purchasing scenarios.
• Exception-based processing, so finance teams could focus on transactions that required judgement.
• Process standardisation over excessive customisation, adapting business processes where possible and not recreating every historical exception in the new solution.
• Phased implementation, delivering value quickly while more complex scenarios were introduced progressively.

The program had joint sponsorship from Finance and IT. The CFO and CIO organisations aligned around a target go-live date and maintained that commitment throughout implementation.

The solution

SpendConsole was implemented as an AI-powered orchestration layer for Shell's Accounts Payable operations, integrating with SAP and progressively automating the invoice lifecycle. The implementation was deliberately phased instead of attempting to transform every scenario simultaneously.

Phase 1 - Non-PO

The first workflows focused on non-PO scenarios, establishing the operating model, integrations and approval processes around a comparatively straightforward transaction set.

Phase 2 - Goods

PO-based goods scenarios followed, incorporating purchasing and goods receipt information into automated validation and matching.

Phase 3 - Services

Service invoices came next, bringing more complex requirements — including Service Entry Sheets and service-specific validation — into the automated process.

Across these scenarios, SpendConsole brought together:

• AI-powered invoice and document capture
  
• Automated invoice entry
  
• Purchase Order matching
  
• Goods receipt validation
  
• Service Entry Sheet validation
  
• Supporting-document processing
  
• Tax and business-rule validation
  
• Approval workflow orchestration
  
• Exception management
  
• SAP integration
  
• End-to-end invoice visibility
  
• Digital audit trails

Most invoices moved through capture, matching, validation and posting without anyone touching them. Finance stepped in only where the platform flagged an exception; a missing document, a mismatched PO, or a validation that needed a judgement call.

A 12-week transformation

The implementation showed that enterprise finance transformation does not require a multi-year program. By breaking the problem into manageable business scenarios, Shell and SpendConsole progressively expanded the solution while maintaining momentum toward the overall transformation objective.

Initial scenarios were live within approximately eight weeks. Goods scenarios followed at approximately ten weeks, and more complex services scenarios at approximately twelve weeks.

Results

Within weeks of go-live, Shell began seeing meaningful changes in the way Accounts Payable operated.

• 50–70% productivity improvement
  Activities that previously consumed significant Finance capacity could increasingly be performed automatically by the platform.
• Processing from approximately 5 minutes to under 1 minute
  Invoices that previously required around five minutes of manual processing could be handled in under a minute in automated scenarios.
• End-to-end visibility
  Finance gained visibility across the invoice lifecycle, instead of relying on fragmented communication and manual follow-up to understand where an invoice was sitting.
• Less repetitive work
  Invoice entry, document processing, validations, matching and workflow activities could increasingly happen automatically, allowing finance teams to focus on exceptions and higher-value activities.
• Stronger process discipline
  Automation surfaced master-data and process issues that had previously been absorbed through manual work, creating an opportunity to improve supplier data, standardise business rules and simplify processes.
• A platform for further innovation
  The team began identifying other processes where AI, orchestration and automation could create value, instead of treating automation as a single Accounts Payable project.

What made the transformation work

The Shell implementation relied on joint finance and IT ownership. The program benefited from pragmatic leadership across finance, operations and technology, with teams working collaboratively toward a shared go-live objective.

The teams challenged existing ways of working and adopted standardised approaches wherever practical, instead of attempting to reproduce every historic process through customisation.

When issues emerged, from master data to SAP customisations and complex business scenarios — the focus stayed on solving them together without losing sight of the target date. That combination of sponsorship, pragmatism and agility took the transformation from implementation to live operations in approximately 12 weeks.