For millions of Africans, linking a bank account to a budgeting app still feels like a leap of faith. Open banking frameworks exist in fewer than half of the continent’s countries, and repeated data breaches have left consumers wary of sharing financial credentials. Budge AI, a start-up launching this week on iOS and Android, believes it has found a way around that trust deficit — by never asking for a bank password.
The app reads the transactional SMS and email alerts that users already receive from banks, mobile wallets and payment platforms, then applies natural language processing to extract amounts, merchants and spending categories. No bank login. No API. No manual spreadsheets. “We realised the data people need is already landing in their pockets every day,” said chief executive Eloke Ikiliagwu. “It just isn’t being structured or interpreted.”
Ikiliagwu co-founded the company with Andrew Glago, who leads product development. Both have backgrounds in software engineering and product design. They built Budge AI after watching friends and family ignore a stream of transaction notifications while struggling to maintain paper budgets. The founders argue that in markets where open banking is embryonic and trust in third-party account access is low, the message-inbox approach is a pragmatic alternative.
Extraction and deduplication
Budge AI’s engine is trained on notification templates from more than 1,500 financial institutions and payment platforms. It deduplicates entries that appear in both SMS and email, then assigns categories. Users can correct mislabelled transactions, feeding back into the model. The process requires no direct bank integration or screen-scraping — a design choice that is as much a necessity as a feature in a region where even large lenders may lack API infrastructure.
The app’s standout feature is “Ask Budge”, a conversational AI interface. Users can type or speak queries such as “How much did I spend on transport last week?” or “Can I afford a new fridge this month?” The system draws on categorised history and income patterns to generate data-driven answers, with disclaimers reminding users that it is not financial advice. In a brief demonstration, the feature returned a breakdown of transport spending in seconds, flagging a 40% month-on-month jump in ride-hailing costs.
“Most budgeting apps show you charts and expect you to figure it out,” Glago said. “We’re letting people ask the question they actually have, in the moment. It feels less like accounting and more like texting a friend who happens to know where your money went.”
The notification lottery
Financial alerts are far from uniform. Some banks send an SMS for every transaction; others only notify above certain thresholds or omit merchant names. Email receipts from informal vendors may use non-standard templates or local languages. Deduplication errors can lead to double-counting. Ikiliagwu acknowledged the challenge: “No automated extraction is perfect. But when users correct a transaction, that feeds back into the model. The system gets sharper with use.”
Cash also remains a significant blind spot. A large share of transactions in many African economies happens in physical currency, and Budge AI only captures those if a user manually enters them. Cryptocurrency transactions are not supported.
Asking for permission to scan SMS and email inboxes is, however, a delicate request. The company says as much processing as possible happens on-device, with only encrypted, anonymised data sent to the cloud for the conversational AI function. It has secured data protection clearances from regulators in a key African market — the company declines to name it, citing commercial sensitivity — confirming that the clearance covers the handling of financial personal data under local law.
Where open banking left a gap
Global personal finance apps like Emma, Spendee and Wallet have users across Africa, but they typically rely on open-banking APIs or manual bank logins. Mobile money services such as M-Pesa offer in-app statements, while local start-ups have built USSD-based trackers. The common thread is that almost all require direct account access or manual data entry. Budge AI’s differentiators are threefold: no bank login, automated extraction from message alerts, and a conversational AI that lets users query their spending in plain language.
“We’re not asking people to trust us with their bank credentials or to set aside time to input receipts,” Glago said. “We’re meeting them where they already are, in their message inbox.” The app tracks bank transfers, card payments, digital wallets, invoices and email receipts across multiple currencies, stitching together a picture of finances that reflects the fragmented way many Africans manage money.
Freemium with a Pro push
Budge AI operates on a freemium model. The free tier covers basic transaction tracking and categorisation, while a Pro tier unlocks premium features — extended transaction history, unlimited budget creation and custom categories. The company is still testing pricing in select markets and has not disclosed the fee. Ikiliagwu said user growth and data quality, which are already gaining momentum, are the immediate priority. “We believe that if we deliver genuine insight, a subset of users will pay for deeper functionality.”
Whether Budge AI can scale beyond early adopters hinges on execution. For now, the start-up is betting that removing the setup friction will unlock a habit that traditional personal finance apps have failed to instil — and that Africa’s mobile-first, notification-rich environment is the right place to test that bet.