ClementIQ

Inspiration Having savings doesn’t always mean being prepared. A balance alone cannot tell someone whether they could handle a layoff, rising bills, or an unexpected repair.

We built ClementIQ to answer three questions: Where do I stand? What happens if life changes? What can I do next? What it does ClementIQ turns the challenge’s anonymized banking data into a personalised financial resilience plan.

Know the customer: see typical income, spending, fixed costs, savings buffer, and financial runway.
Test a shock: simulate job loss, higher household bills, unexpected expenses, or combined scenarios.
Make a plan: explore an emergency-fund goal and a backup plan based on the customer’s own numbers. In our demonstration, account #20133 has approximately $30,198 in savings. Under the challenge’s layoff scenario, with income replaced at 55%, the account faces a monthly shortfall of approximately $7,169. Its buffer lasts just 4.2 months.

ClementIQ makes that consequence visible, then helps the customer explore possible responses. How we built it We built the interface with Next.js, React, and TypeScript, with charts that show how a customer’s financial position changes under different scenarios.

Our backend uses Python, FastAPI, and pandas to process the banking data and calculate customer metrics, shock impacts, and recovery plans.

We separated the calculations from the explanations:

Clement Core calculates the financial results using deterministic formulas.
Clement Gemini uses the Google Gemini API to explain those results in plain language. The financial calculations follow the challenge’s definitions, including median monthly income and spending, and a savings buffer based on the last six monthly average balances. Challenges we faced One challenge was keeping the numbers consistent across the dashboard, simulator, and recovery plans. The same account and assumptions needed to produce the same results throughout the experience.

Another was interpreting incomplete transaction labels. An unlabelled withdrawal is not necessarily optional spending, so suggested reductions must be presented as possibilities to review—not guaranteed savings.

We also needed to distinguish between preparing before a shock and responding after it. Our Goal Plan uses the customer’s baseline finances to build savings, while the Backup Plan addresses the simulated shortfall. What we learned Financial resilience depends on the relationship between income, spending, and savings—not simply the size of an account balance.

We also learned that useful AI needs a clear role. Keeping financial calculations in code and using Gemini for explanations makes the results easier to check and understand.

Most importantly, a warning becomes more useful when it comes with a practical next step. What’s next We want to improve expense classification, validate the planning assumptions further, and help customers compare more realistic recovery options.

Our goal is to make ClementIQ a clearer bridge between understanding financial risk and taking action. Data and acknowledgements Built for the Tangerine: Resilience Rewired track at Sheridan Datathon 2026, using the supplied PKDD’99 Discovery Challenge Financial dataset from the CTU Prague Relational Dataset Repository.

ClementIQ is an educational prototype, not financial advice. The demonstrated EI scenario follows the challenge’s simplified assumptions.