FinChief

• 

  First page

Inspiration

Most personal-finance apps explain where money has already gone. I wanted to build something that helps people pause before a purchase and understand the trade-offs clearly, without affecting their credit score or pretending to replace a financial adviser.

What FinChief does

FinChief is a personal financial decision-support app for one-off purchases. A user enters a fictional household position, including income, unavoidable commitments, general savings, an emergency fund and goal savings. They then enter a proposed purchase and choose exactly one funding source.

FinChief calculates whether the purchase is fully funded, how much of the selected savings pool it consumes, what remains afterwards, whether emergency resilience changes, how monthly cash position is affected, and whether a financial goal is delayed. It also shows a 12-month liquid-savings projection and several transparent alternatives such as reducing the purchase amount, waiting and saving, or choosing a different funding source.

The app deliberately keeps general savings, emergency savings and goal savings separate. It also separates the household's baseline financial position from the incremental effect of the proposed purchase, so an existing weakness is not incorrectly attributed to the decision.

How I built it

I built FinChief in Codex using GPT-5.6. I worked through the product as a sequence of user stories and reviewed each phase before moving on. Codex accelerated the architecture, implementation, test creation, documentation and UI refinement. I corrected important assumptions during development, especially around savings-pool separation, emergency-fund coverage, goal delay and opportunity-cost wording.

The financial engine is deterministic and uses Python Decimal arithmetic. Python calculates every financial figure and classification. The optional GPT-5.6 explanation layer can only select from approved calculated facts and provide plain-language framing. Python then validates the response and reinserts the authoritative numeric statements. If GPT is unavailable or the response fails validation, FinChief automatically falls back to a deterministic explanation.

Technical highlights

• Responsive, mobile-first Streamlit interface
• Deterministic Decimal-based financial engine
• Separate general, emergency and goal savings pools
• Baseline-versus-decision impact reporting
• 12-month liquid-savings projection
• Deterministic alternatives module
• Safeguarded optional GPT-5.6 explanation layer
• 62 automated tests
• No credit check and no bank connection

What I learned

The most important lesson was that a finance app must distinguish affordability from resilience and opportunity cost. A purchase can be fully funded while still creating a meaningful trade-off. I also learned that AI-generated explanations are safer and more trustworthy when the model is not allowed to invent or calculate the underlying facts.

Important note

FinChief is an educational prototype that uses fictional demonstration data. It does not provide regulated financial, mortgage, tax or investment advice.

Built With

• altair
• gpt-5.6
• openai-codex
• pytest
• python
• streamlit