Set to Make Your Ambitions a Reality?
The Challenge
Early 2025, a fintech startup applied for a £60K loan from CTM but was rejected due to limited financial history & first-time founder & misunderstood revenue forecast metrics. 3 months later, the startup secured $700K VC funding. CTM got criticised in startup circle, causing reputation damage despite wasted resources for the previous credit analysis. 3 problems were raised:
- Systemic blind spots: 60% of startup applicants got rejected by outdated scoring methods.
- Competitive pressure: Their startup-lending unit underperformed for 3 consecutive quarters, losing share to AI-enabled fintechs.
- Rising risk: Ironically, several approved SMEs defaulted within 6 months based on tax data the system was built to trust.
The Solution
Technical Approach: CTM partnered with a regional AI consultancy to develop StartLens, an intelligent credit risk system that understands modern startups.
Key Components:
- Alternative data fusion: Leverages non-traditional data sources—founder backgrounds, business plans, user growth metrics, and live platform data (e.g., Stripe, Xero).
- Reinforcement learning: Learns from credit analysts’ overrides to refine decision logic in real time.
- Smart escalation: Flags high-uncertainty cases for human review, with AI-generated summaries.
- Regulatory guardrails: Integrated fairness and compliance checks to prevent bias and ensure transparency.
- Embedded APIs: Seamless integrations with business platforms for real-time performance data.
Implementation Roadmap
1 - POC & Planing
Define and draft solution roadmap in 10 weeks.
2 - Data integration
Establish API connections in 8 weeks.
3 - API implementation
Deploy embedded APIs for real-time data ingestion and decision outputs in 8 weeks.
4 - Rollout A/B testing
Control Group: 50% of applications reviewed + Test Group: 50% processed.
Results
AI playbook for BSFI leaders
Explore high-impact use cases to inspire your AI adoption
