New AI Feature Scans Printed and Handwritten Receipts to Fill in Till and PayBill Details
Paying through M-PESA is part of everyday life in Kenya, from buying groceries at a neighbourhood shop to settling bills at hardware stores and service businesses. However, entering a long Till or PayBill number manually can take time and create opportunities for mistakes, particularly when customers are paying in a hurry. Safaricom’s AI Pay feature aims to make this process easier by using artificial intelligence to read payment details from receipts.
AI Pay is available through Safaricom’s My OneApp experience, allowing customers to scan a printed or handwritten receipt instead of entering every detail themselves. When the information is sufficiently clear, the feature can identify the relevant Till or PayBill number and payment amount, then populate the appropriate fields on the payment screen. Customers can review the information before authorising the transaction.
The development reflects the growing role of artificial intelligence in digital financial services. Rather than requiring customers to learn a new payment method, AI Pay works within the familiar M-PESA payment process. This approach could make everyday transactions more convenient for people who regularly pay merchants using details printed or written on receipts.
Safaricom says My OneApp has recorded more than seven million downloads, giving the feature access to an established customer base. Its practical value, however, will depend on how accurately it reads different types of receipts and how easily customers can verify the information before completing payments.
The feature also addresses a familiar challenge in Kenya’s retail economy. Many small businesses continue to issue handwritten receipts and invoices, while others use printed documents with varying layouts. By supporting both formats, AI Pay is designed to accommodate a range of everyday payment situations rather than relying exclusively on standardised documents.
How Safaricom AI Pay Works
AI Pay uses receipt scanning to reduce the amount of information customers must enter manually when making eligible M-PESA payments.
Instead of repeatedly checking a receipt and typing its details into the app, a customer can scan or share the payment information through the supported experience. The system then analyses the document and identifies details such as the merchant’s Till or PayBill number and the amount to be paid.
When the information is recognised successfully, the app fills in the relevant payment fields. The customer can then check the details and complete the transaction using the usual M-PESA authorisation process.
Safaricom Group Chief Technology Officer James Maitai described the feature as a way for the system to process receipts, identify the relevant payment information and determine the amount required.
The final approval remains with the customer. Depending on the available device and app options, the person paying can enter an M-PESA PIN or use fingerprint authentication to authorise the transaction.
This distinction is important because automated receipt recognition does not replace the customer’s responsibility to verify the payment details. A receipt may contain unclear handwriting, outdated information or details that the system cannot interpret correctly. Checking the populated fields before approval remains an essential step.
Benefits of AI Pay for M-PESA Customers
Faster Till and PayBill Transactions
One of the most immediate potential benefits is reduced payment preparation time. Customers no longer need to type every digit manually when the system successfully recognises the receipt.
Although entering a payment number may take only a short time, repeated transactions can make the process inconvenient. The difference may be particularly noticeable at busy retail counters, where customers want to complete purchases without delaying others.
AI Pay could also help people who frequently make payments using invoices or receipts supplied by service providers and merchants.
Fewer Manual Entry Errors
Manually entering payment details creates a risk of mistyping numbers. Even a small mistake can cause a transaction to fail or direct a customer towards an unintended payment destination.
By extracting information from a receipt, AI Pay reduces the need for repeated manual entry. However, scanning cannot guarantee that every detail will be interpreted correctly. Customers should confirm that the displayed merchant number and amount match the information provided before authorising payment.
This verification step remains important even when the app fills in the fields automatically.
Support for Handwritten Receipts
The ability to recognise handwritten documents could make the feature useful across a wide range of Kenyan businesses.
Small shops, hardware outlets, repair services and other traders may issue receipts or invoices written by hand. A tool that can interpret legible handwriting could help customers make payments without requiring merchants to change their existing processes.
Results may vary depending on handwriting quality, lighting, paper condition and the clarity of the payment information. Faint printing, damaged receipts and poorly written numbers may still require customers to enter some details themselves.
A Familiar Payment Experience
AI Pay operates within the broader My OneApp and M-PESA experience rather than introducing a completely separate payment system.
Customers can therefore use receipt recognition as an additional convenience while retaining the familiar authorisation process. For people already comfortable with M-PESA, this could make the feature easier to adopt.
What Customers Should Check Before Paying
Automation can simplify a payment, but it does not remove the need for caution. Customers should review the merchant details and payment amount after scanning a receipt, especially when dealing with handwritten numbers or unfamiliar businesses.
If the information displayed on the screen differs from the receipt, customers should correct it before proceeding. They should also avoid authorising a transaction when the merchant details are unclear or cannot be independently verified.
Customers should never share their M-PESA PIN with another person to complete a payment. They should use only the official app experience and the normal security controls provided for their account.
The feature may not work equally well with every receipt. Where the system cannot identify the required information, manual entry may still be necessary. Its usefulness will therefore depend on both the quality of the document and the accuracy of the recognition process.
How AI Pay Fits Into Safaricom’s Digital Strategy
Safaricom’s introduction of AI Pay reflects a broader effort to incorporate artificial intelligence into services customers already use.
My OneApp provides a platform for adding features to an existing digital ecosystem. With the app reportedly exceeding seven million downloads, Safaricom has an established audience for testing and expanding new customer-facing services.
For merchants, receipt scanning could help preserve familiar ways of issuing payment information while making it easier for customers to complete transactions. For consumers, the feature offers another way to reduce repetitive steps during everyday purchases.
Its long-term impact will depend on practical performance. Recognition accuracy, ease of use, compatibility with different receipt formats and the clarity of payment verification will all influence whether customers continue using it.
What AI Pay Means for the Future of M-PESA
Safaricom AI Pay introduces a more automated approach to preparing Till and PayBill payments. By recognising details on printed and handwritten receipts, the feature aims to reduce manual typing and make routine transactions more convenient.
The service does not eliminate the need to check payment information, and unclear receipts may still require manual input. Nevertheless, integrating receipt scanning into the familiar M-PESA experience could offer a useful improvement for customers and merchants who handle frequent digital payments.
As AI-powered features become more common in financial applications, the key measure of success will be whether they make transactions simpler without compromising accuracy, transparency or customer control.
