Singapore Soft-Launches InvoiceNow AI Skills Library to Support Automated E-Invoicing

Singapore’s IMDA has soft-launched the InvoiceNow AI Skills Library, offering reusable AI workflows for invoice ingestion, Peppol packaging and 3-way matching. The initiative supports businesses preparing for wider GST InvoiceNow adoption and aims to reduce manual invoice-processing work.

Singapore’s Infocomm Media Development Authority (IMDA) has soft-launched the InvoiceNow AI Skills Library, a new reference resource designed to help businesses and solution providers apply artificial intelligence to common invoicing processes. The initiative combines AI-based workflow examples with InvoiceNow’s structured e-invoicing environment, giving businesses practical starting points for automating invoice capture, preparation, and matching.

InvoiceNow is Singapore’s nationwide e-invoicing network and is based on the international Peppol framework. It enables invoices and other supported business documents to move directly between accounting, finance, and enterprise systems in structured, machine-readable formats. Unlike invoices exchanged only as PDFs or paper documents, structured InvoiceNow data can be processed by software without requiring the same level of manual data entry.

This structured foundation is particularly relevant for AI. AI tools generally perform more effectively when they operate on consistent data and clearly defined workflows. IMDA’s new library is intended to demonstrate how businesses can combine the standardized InvoiceNow environment with reusable AI instructions and workflow components rather than developing every automation process from the beginning.

The first version of the InvoiceNow AI Skills Library focuses on three practical invoicing use cases.

The first is incoming invoice ingestion. This skill is intended to extract information from supplier invoices and convert it into structured data that can be ingested into the buyer’s system. For businesses that continue to receive invoices in different formats, this could reduce the amount of manual capture required before an invoice enters the accounting or accounts-payable process.

The second skill covers outgoing invoice packaging. It converts invoice information into the required Peppol format so that the invoice can be transmitted to the buyer through InvoiceNow and, where applicable, the relevant invoice information can also be submitted to the Inland Revenue Authority of Singapore (IRAS). This is particularly relevant as Singapore progressively expands the GST InvoiceNow Requirement.

The third use case is 3-way matching. This process compares an invoice with the corresponding purchase order and goods receipt. Matching these records can help determine whether the invoiced amount, ordered goods, and received goods are consistent. When differences are detected, the transaction can be directed for exception handling instead of being processed automatically.

Importantly, the library does not create a new mandatory AI requirement. IMDA presents it as a practical reference containing AI Skills and suggested prompts that businesses can adapt to their own processes, software and operating requirements. Its terms of use also make clear that businesses remain responsible for assessing whether the examples are appropriate and for carrying out their own due diligence before implementing or modifying them.

Connection With the GST InvoiceNow Mandate

The development should be viewed alongside Singapore’s wider GST InvoiceNow programme. IRAS requires affected GST-registered businesses to submit invoice data using InvoiceNow-Ready Solutions, with implementation taking place in phases. Newly incorporated companies voluntarily registering for GST have been subject to the requirement since 1 November 2025, while all new voluntary GST registrants have been covered since 1 April 2026.

The mandate will expand further from 1 April 2028. New compulsory GST registrants and existing GST-registered businesses with total annual supplies of no more than S$200,000 will be included first. Existing businesses with supplies of up to S$1 million follow in April 2029, those with supplies of up to S$4 million in April 2030, and businesses above S$4 million in April 2031. IRAS states that all remaining GST-registered businesses will therefore be brought into the requirement by April 2031.

For software providers, ERP developers and larger enterprises operating their own finance systems, the AI Skills Library could provide another implementation layer around the existing InvoiceNow technical framework. IMDA already provides technical specifications covering overall system design, data extraction and transformation, and Access Point services for GST InvoiceNow. The AI library does not replace these specifications. Instead, it can be used to explore automation around processes that ultimately need to produce or consume compliant structured invoice data.

Retail businesses may also find the development relevant even where most customer-facing transactions originate in POS systems. IRAS guidance confirms that invoice data submitted under the GST InvoiceNow Requirement can include sales invoices, tax invoices, simplified tax invoices, serially numbered receipts, debit notes and credit notes. It also allows certain supplies data from POS systems and simplified tax invoices to be aggregated before submission to IRAS.

This means the immediate impact of the AI Skills Library is expected mainly in back-office processes rather than at the checkout. Retailers with significant supplier volumes, purchase orders and goods-receipt processes may be able to use AI-supported invoice ingestion and 3-way matching to reduce manual work in accounts payable. Retailers issuing B2B invoices may also benefit from automated preparation of structured InvoiceNow documents.

Businesses should nevertheless distinguish between AI automation and regulatory compliance. Using an AI Skill does not by itself demonstrate compliance with GST InvoiceNow requirements. Businesses still need an appropriate InvoiceNow-Ready solution, connectivity through the InvoiceNow network, correct data mapping and controls ensuring that information submitted to IRAS is complete and accurate. IMDA requires InvoiceNow-Ready Solution Providers to support the necessary InvoiceNow and GST InvoiceNow functionality through accredited Access Points.

The soft launch therefore represents an expansion of InvoiceNow from a structured e-invoicing network toward a broader automation environment. For businesses already preparing for mandatory GST InvoiceNow implementation, the new library provides an opportunity to assess whether the same structured data required for compliant electronic invoicing can also support more automated finance workflows.

How will the InvoiceNow AI Skills Library impact retail businesses?

The immediate impact is expected mainly in finance and back-office operations, where retailers may use the Skills to automate supplier invoice ingestion, 3-way matching, and preparation of structured B2B invoices. Retailers should treat the AI library as an optional automation resource and continue preparing separately for their applicable GST InvoiceNow compliance deadline.

Here you can find the link to AI library. Source

Ljubica Blagojević, Junior Legal Research Specialist at Fiscal Solutions

Linkedin profile