All projects PRODUCT FACULTY

Navi.

By Aida Michel, Raj, Tatiana, Chiheb & Nidhi

Loading video…

Navi is a voice-first AI budgeting coach built for users that traditional fintech apps systematically fail: migrants, blue-collar workers, young professionals, and the financially anxious. Instead of dashboards, forms, and category trees, users just speak naturally — *"I spent 12 dirhams on breakfast"* — or snap a receipt or have Navi read SMS transaction alerts. The AI processes the input, categorizes it, logs it, and replies in plain language.

Designed for low-literacy, hands-free use, Navi explicitly frames itself as a budgeting assistant, not a financial advisor — a deliberate scope choice that keeps it safe to use in regulated regions and across the wide variance of migrant financial contexts (no country assumptions, no embedded credit advice).

Core AI capabilities: voice expense logging, receipt OCR, SMS expense extraction, AI categorization of messy transactions (e.g. a Walmart purchase split across categories), real-time spending nudges, bill reminders, affordability checks ("Can I afford this?"), subscription detection, drift alerts, and goal impact previews — all routed through a single strict JSON output contract that prevents the model from drifting into financial advice.

Powered by GPT-5o-mini (chosen for the cost-latency-safety balance against the conversational-quality requirement), with Supabase persistence and trust-by-design scaffolding: continuous bias checks, transparent consent flows, and migrant-safe defaults. B2C-first, with a B2B2C secondary engagement model for employers and remittance partners.