Weidai \China

Step 2 - Fraud Detection and Model Customization (Validation): Add real-time fraud detection layer using device fingerprinting, velocity checks, and anomaly detection. Build a simple dashboard for lenders to monitor model performance, see feature importance, and flag suspicious applications. Introduce model customization: allow lenders to adjust risk thresholds and incorporate their proprietary data. Expand to 5-10 customers across 2 markets. Introduce SaaS tier ($2,000-5,000/month) for dashboard access plus per-API-call pricing. Goal: $50K MRR, 100,000 monthly credit decisions, demonstrate 25-30 percent reduction in fraud losses for at least one customer. Validate that customers will pay for both API access and analytics.

Step 3 - Full Platform with Portfolio Management (Growth): Build out the full CreditOS platform: automated collections workflows (SMS/email campaigns, payment reminders), portfolio monitoring (cohort analysis, vintage curves, early warning alerts), and A/B testing framework for credit policies. Integrate with core banking systems (Mambu, Temenos) and loan management systems. Expand to 20-30 customers across Southeast Asia and one additional region (Latin America or Africa). Introduce enterprise tier ($10,000-25,000/month) with dedicated support and custom model development. Goal: $500K MRR, 1M monthly credit decisions, sign at least 2 top-10 digital banks in target markets. This is the inflection point where network effects kick in—more customers mean more data, better models, and stronger moat.

Step 4 - Data Network and Vertical Expansion (Moat): Launch CreditOS Data Cooperative: customers can opt into anonymized data sharing to improve model performance across the network (with privacy controls and incentives). This creates a data moat—the more lenders use CreditOS, the better the models become. Expand into adjacent verticals: BNPL providers (Affirm-style), SMB lending (invoice financing, working capital), and embedded finance (e-commerce platforms, gig economy apps). Build vertical-specific models and workflows. Introduce a marketplace for third-party data providers (telco data, utility payments, social data) to plug into CreditOS. Goal: $3M ARR, 50-100 enterprise customers, process 10M+ monthly credit decisions. At this scale, CreditOS becomes infrastructure—switching costs are high, and the data network effect creates a sustainable competitive advantage.