Fintech · Decision intelligence
Klariq: Choosing the right credit card before you pay
Klariq helps you decide which card to use before you pay. It identifies the merchant, evaluates cards in the simulated wallet, and compares estimated value while keeping uncertain reward coding visible.
- Domain
- Fintech · Decision intelligence
- Role
- Product designer · UX engineer
- Platform
- Mobile web
- Year
- 2026
- Scope
- Product design, interaction model, reward logic presentation, prototype, and implementation hardening
- Validation
- 512 / 512 assertions
30 Second Summary
Problem
Credit card advice breaks down when merchant identity, reward coding, or purchase context is unclear before the math even starts.
Role / Scope
I designed the product, interaction model, reward logic presentation, prototype, and the checks that keep the implementation consistent.
System
Merchant understanding → card rules → normalized value → recommendation, with clarification only when it can change the winner.
Outcome
A working financial decision prototype backed by 512 implementation assertions and separate Chromium QA.
Decision System
- Understand merchant. Identify what the merchant appears to be and surface uncertainty when it matters.
- Evaluate card rules. Apply the relevant earning rules, caps, channels, and activation state.
- Normalize and rank. Convert unlike reward currencies into estimated value so cards can be compared on the same basis.
- Explain the choice. Show the recommendation, alternatives, and evidence without hiding uncertainty.
Primary decision flow
The main flow keeps merchant analysis, card evaluation, ranking, and explanation visible as separate steps so the recommendation does not feel like a black box.
Recovering from merchant uncertainty
When a merchant such as Velora is ambiguous, Klariq asks the user to describe it only when that context can change which card wins.
Financial reasoning
Point Valuation changes the estimated monetary value of points and miles without changing the raw earning rate. Recommendation Style makes the ranking logic explicit rather than silently changing the answer.
Track and learn
Track Purchase confirms the save and returns to Home, where the tracked purchase appears as a non-interactive Pending status. Later verification is a separate contextual step that can add evidence without rewriting the original decision.
What changed
The final prototype makes reward logic explicit, asks fewer but more useful questions, protects historical context, and keeps implementation validation separate from browser QA.
