At traditional banks and financial providers, risk analysis is often a laborious manual process that can take several days or even weeks. Factors such as age, business type, and location also play a role, leading to negative influences on the credit decision. However, these factors are difficult to justify, especially for online companies. Fulfin, on the other hand, uses an AI-supported algorithm to calculate creditworthiness.
This is based on transaction data from the digital account view and does not take into account any potentially discriminatory Variables. AI models and statistical methods such as NLP, LLMs, tree-based ML modeling, and SHAP values are utilized throughout the entire lending process. According to the company, the startup invested around 4 million in the development of its AI solution. The result is a proprietary financing platform that enables fundamentally more inclusive lending, instantly and without human bias.
With research funding of 1.1 million euros, the Federal Ministry of Research and Education is now honoring Fulfin's AI developments. The startup has also received the BSFZ seal of approval from the Research Allowance Certification Office, which further recognizes the pioneering nature of the in-house developed product innovation in the field of AI-supported risk assessment.
“Fulfin challenges the status quo in lending”
Samarth Mehrotra, Head of Data & AI at Fulfin, emphasizes:
"Lenders often make credit decisions based on the 'probability of default' (PD). The underlying PD values typically come from credit bureaus and are often biased against young, fast-growing companies because they rely on static data sources such as financial statements. By developing a new decision-making engine in-house based on real-time open banking data, Fulfin is challenging the status quo in lending."
While traditional lending institutions still rely on the experience of their risk managers for their decisions, Fulfin gives them a new role. They transform from "decision makers" to "decision reviewers" who control the algorithm's outcome. This allows a larger number of loan applications to be processed and enables more cost-effective lending for SMEs.
