FamilyFreely

Inspiration

Household money challenges are not always calculation problems. They often arise because family members understand income, spending, responsibilities, and progress differently. We wanted to create a private, shared workspace that improves transparency without placing one person in the role of “money police” or exposing children to punitive financial controls.

What it does

FamilyFreely is a responsive household finance PWA for managing income, budgets, transactions, accounts, goals, recurring finances, imports, and shared decisions. Its family-safe Points & Rewards system allows adults to define observable positive behaviours, while children receive an age-appropriate points balance and private history. Families can create reward catalogues, request and approve redemptions, reverse mistakes, and ask questions about individual point entries. Basic needs cannot be offered as rewards. Response cost is disabled by default. If a family chooses to enable it, the feature requires separate permission, guardian acknowledgement, predefined reasons, fixed limits, a zero-balance floor, and an immutable audit trail. FamilyFreely also supports recurring income and bills, review-first CSV imports, offline-safe transaction capture, role-based permissions, account activity history, and human-reviewed AI assistance.

How we built it

The frontend is a responsive Next.js and TypeScript PWA designed for phones, tablets, desktops, and installed use. A Fastify API runs on Cloudflare, while PostgreSQL stores financial records, household permissions, immutable point transactions, recurring schedules, imports, and audit history. Offline financial writes use IndexedDB and idempotency keys to prevent duplicate transactions. Database changes use additive migrations to preserve existing household data. GPT-5.6 is used through the OpenAI Responses API with strict structured outputs, minimized inputs, disabled storage, and hashed safety identifiers. The AI can draft positive behaviour and reward ideas or explain product workflows. However, it cannot change money, permissions, points, rewards, or account state.

Challenges

The central challenge was designing a points system that supports positive reinforcement without enabling arbitrary punishment. At the same time, we had to preserve authorization across household roles, maintain immutable histories, prevent duplicate financial writes, protect existing data during migrations, and keep AI useful without giving it decision-making authority.

Accomplishments

We delivered a coherent cross-device application with family-safe Points & Rewards, immutable reversals, child questioning workflows, recurring finances, review-first imports, offline reconciliation, role-aware authorization, guarded deletion, accessibility improvements, and live GPT-5.6 guidance.

What we learned

Family-safe design requires governance, not only friendly interfaces. Permissions, limits, explanations, reversibility, and auditability must be implemented consistently across the frontend, API, and database. This means that safety cannot be added as a final interface layer. We also learned that AI is most appropriate here as a structured drafting and explanation tool. Every consequential decision remains with the family.

What’s next

Next, we plan to conduct structured pilots with consenting families, broaden localization, strengthen notification delivery, undertake further accessibility testing, and prepare a separately reviewed production rollout.