Personal finance app

SpendWise

Financial clarity without the complexity.

SpendWise reads the purchase alerts a bank already sends, understands what each one means and keeps an accurate running total — then helps people see where their money is going.

Ownership: A WianTribe product
Year: 2026
Platform: Android
Status: In development

SpendWise Insights screen showing average purchase and a period-end estimate

SpendWise home screen showing total spent against the monthly budget

Overview

Our role: Product strategy
UI/UX design
Mobile engineering
Data and privacy architecture
Disciplines: Application Development
Product Design
UI/UX
Product Engineering

Most people already receive a notification every time they tap their card. SpendWise turns those alerts into structured transactions, fills in history from downloaded statements and presents the month clearly: what has been spent, what remains and what has changed.

SpendWise is a WianTribe product in active development for Android. We designed the experience and engineered everything beneath it, from the notification listener to the parsing engine and reporting.

The challenge

Accurate numbers are harder than they look.

Budgeting tools often ask for online banking credentials or rely on third-party data connections. SpendWise works from information that is already on the phone instead.

That creates its own challenge. One alert can mention both a purchase and a decline; a bill payment is spending while a credit card payment isn't; and the same purchase can arrive twice, then again weeks later on a statement. Getting any of this wrong quietly invents spending that never happened.

Our approach

Only purchases increase spending.

That single product rule shapes the whole system. Refunds, card payments, transfers, declines and duplicate alerts are recognized and kept out of the total — and when the app is unsure, it takes the conservative reading.

Read-only by construction
SpendWise has no payment capability and never stores banking passwords, PINs, CVVs or full card numbers.
Honest insights
Comparisons use matching periods, estimates are labelled as estimates and percentage changes are hidden when there's nothing meaningful to compare.
Preview before saving
Statement imports show exactly what will be added, merged or skipped before anything is written.

SpendWise home screen with total spent, daily budget left and daily average — Fig. 01Home, Insights, Budgets and dark mode — rendered from the app's own interface using fictional sample data.

SpendWise Insights screen comparing this period with the prior period — Fig. 01Home, Insights, Budgets and dark mode — rendered from the app's own interface using fictional sample data.

Key capabilities

A clearer view of every month.

01
Automatic capture
Reads purchase notifications from selected banking apps in the background, starting with CIBC and Wealthsimple.
02
Statement import
Imports CSV and PDF statements to fill in history, reconciled against transactions that were already captured.
03
Spending insights
Daily spending, category breakdowns, top merchants, six-period history and like-for-like comparisons.
04
Budgets with guidance
Overall and category budgets with a daily allowance, plus optional suggestions based on previous spending.
05
Ask SpendWise
Answers questions about spending with a private query engine. Financial data is not sent to an external AI provider.
06
Financial Vault
Keeps the statements behind imported history, so source documents are there whenever they're needed.

Engineering

Correctness, designed in.

Every number SpendWise shows has to be right — so accuracy is built into the architecture and tested at every layer.

01
A parser built on product rules
Alerts are classified through an ordered chain of rules, most specific first, so “could not be completed” is recognized as a decline and a utility bill still counts as spending.
02
Duplicates caught, repeats kept
A transaction's identity and its timing are handled separately, so two copies of one alert merge while the same coffee bought again that afternoon still counts.
03
Every source counted once
Notifications, statements and a bank data connection (when configured) are reconciled, so a purchase is enriched with settled details rather than recorded twice.
04
Privacy in the data layer
Raw notification text is discarded once parsed, and every cloud table is protected with row-level security.
05
Tested at every layer
Hundreds of automated tests cover parsing, reconciliation, reporting and every screen, and run in continuous integration.