The Singapore fintech teaching banks to judge credit risk with explainable AI - and to build a scorecard in a day instead of a year.
In lending, the bottleneck was never capital - it was the risk model. FinbotsAI set out to fix that.
Ask a bank how long it takes to build a new credit scorecard and the honest answer is usually measured in seasons: nine months, sometimes a full year of data wrangling, statistical modelling, validation and sign-off. FinbotsAI, a fintech founded in Singapore in 2017, was built around a blunt observation - that legacy credit systems are too slow and too opaque for the way lending actually moves today.
Its answer is CreditX, a cloud-based, no-code platform that lets a risk team ingest internal, external and alternate data, then automatically build, validate and deploy a custom credit scorecard. The company says the work that once took nine to twelve months can be done in roughly a day. That compression is the product's central claim, and the reason lenders across Asia and the Middle East have signed on.
The founders - Sanjay Uppal, who serves as chief executive, Bala Iyer, and Deeba Kazmi, the chief data scientist - built the platform to serve people who are not necessarily programmers. A risk analyst without Python can still assemble a production model, because CreditX auto-derives more than a hundred variables, handles the feature engineering, and exposes the results through an interface rather than a codebase.
What separates FinbotsAI from a long list of AI-lending hopefuls is less the modelling and more the paperwork around it. Credit decisions are regulated. A model that lifts approvals but can't be explained to a supervisor is a liability. FinbotsAI leaned into that early, becoming among the first fintech solutions globally to complete two Singapore government frameworks - MAS Veritas, which tests for responsible AI in finance, and AI Verify, the world's first AI governance testing toolkit developed by the country's IMDA and PDPC.
The results lenders report are concrete rather than theatrical: north of a 20% increase in approval rates and around a 15% reduction in loss rates, alongside faster model development. In 2022, Accel wrote the company its first outside cheque - a US$3M Series A - to expand the product, the sales effort and the team across offices in Singapore, Hyderabad and Dubai.
"Your trusted AI credit risk platform - approve and lend more, profitably."
CreditX covers the full credit lifecycle. Two companion products extend it into targeting and real-time decisioning.
A no-code AI credit modelling SaaS. Ingests internal, external and alternate data; auto-builds, validates and deploys application, behaviour and collections scorecards with 100+ derived variables, explainability and one-click API deployment.
A propensity modelling product that helps lenders find the customers most likely to respond to - and qualify for - a credit offer, so acquisition spend lands where it converts.
A real-time business-rules and decisioning layer that operationalises scorecards, returning a credit decision in under 0.03 seconds through an API.
Where traditional credit modelling and FinbotsAI part company.
| Dimension | Traditional approach | FinbotsAI / CreditX |
|---|---|---|
| Time to a scorecard | 9-12 months | About a day |
| Who can build it | Specialist data scientists | Risk analysts, no code required |
| Explainability | Often a black box | Explainable, decision-level transparency |
| Regulatory validation | Bespoke, slow | MAS Veritas & AI Verify completed |
| Deployment | Hand-off to engineering | One-click API deployment |
| Decision latency | Batch / minutes | Under 0.03 seconds |
Banks, digital banks, and fintech, SME and microfinance lenders across roughly a dozen markets in Asia and MENA. Named clients include:
FinbotsAI sits at the intersection of AI credit scoring and regtech. It competes with AI-decisioning vendors and incumbents, and against the in-house teams banks would otherwise staff.
Traditional scorecards take the better part of a year. CreditX turns raw data into a deployed model in hours, freeing risk teams to iterate rather than wait.
Regulators and risk committees can't sign off on models they can't read. Explainability and completed governance frameworks make CreditX auditable.
Legacy systems reject applicants with little credit history by default. Using alternate data, CreditX helps lenders responsibly say yes to more of them.
Uppal, Iyer and Kazmi launch FinbotsAI to modernise slow, opaque legacy credit systems.
A no-code AI platform for creating, validating and deploying credit scorecards.
First external capital, from Accel's Fund VII - and KBZ Bank signs on as a client.
Named Best Credit Modelling Solution and completes MAS Veritas and AI Verify.
Partners with Fintech Galaxy to bring CreditX scoring to lenders via FINX Connect.
FinbotsAI licenses CreditX to lenders on a subscription and managed-services basis, adding implementation and model-validation support.
Led by Accel from Fund VII in 2022 - the company's first external investment. Total disclosed funding is around US$4.1M.
Fintech Galaxy (open banking, MENA), plus technology and delivery ties with AWS, IBM, Oracle, TCS and PwC.