The risk team gains autonomy. The institution keeps control.
With uFlow, risk teams design, test, version and update credit policies from a visual editor, without requiring an IT development for every change. The organization keeps permissions, traceability, security and publishing control.
Building the engine in-house gives you full control, but it costs time and a permanent team to maintain it. A generic engine speeds up the start, yet it tends to be rigid and dependent on technical profiles. uFlow strikes the balance: autonomy for the risk team to change policies in hours, within defined permissions and controls, with the governance and traceability a regulated institution requires.
What are the options for automating credit decisions?
What a financial institution gets from each option, across the dimensions that actually matter.
| Build in-houseFull control, but expensive and slow to evolve. | Generic / legacy engineRigid rules, dependent on technical consulting. | uFlowAutonomy for the risk team, with governance and traceability. | |
|---|---|---|---|
| Changing a credit policyFrom the moment it is decided until it runs in production. | Weeks or months (IT release) | Days, requires a technical profile | Hours, self-service for the risk team |
| Autonomy for the business / risk team | Every policy change requires an IT development cycle | Partial, with technical support | Autonomy within defined permissions and controls, with a NoCode editor |
| Policy versioning and rollbackReverting to an earlier version during an incident. | Has to be built and maintained | Limited or manual | Automatic versioning with controlled reactivation of previous versions |
| Testing environments before productionTesting changes without affecting real decisions. | To be built by each team | Depends on the implementation | Built-in testing environments |
| Traceability and audit of every decisionInput, rules applied and outcome, all auditable. | To be built; ongoing maintenance cost | Coverage varies | Complete log + transaction explorer |
| Segregation of roles and permissions | To be built and governed | Available | Per-user permission administration |
| Pre-certified credit bureaus and data sourcesWithout integrating each provider one by one. | Build your own integration per source | Limited catalog | 30+ pre-certified providers across Latin America |
| Machine learning without your own infrastructure | Maintain Python environments | Not covered or limited | Run Python models without operating those environments |
| Certified security | Your own responsibility | Varies by provider | ISO/IEC 27001, 2FA, encryption at rest and in transit |
| Infrastructure and scalability | You provision and maintain everything | Deployment and scalability depend on each provider | Serverless in the cloud, scales automatically |
| Total cost and time to production (TCO) | High capex + a dedicated team | Licenses + consulting | SaaS: typical implementation in weeks, depending on each institution |
| AI agent inside the editorThe agent proposes; the team reviews and publishes. Available in binary and calculation nodes. | You have to build it, and then maintain it. | Depends on the vendor; not always inside the editor. | Builds rules from plain language, audits the order and reads balance sheets. |
Changing a credit policy
From the moment it is decided until it runs in production.
Autonomy for the business / risk team
Policy versioning and rollback
Reverting to an earlier version during an incident.
Testing environments before production
Testing changes without affecting real decisions.
Traceability and audit of every decision
Input, rules applied and outcome, all auditable.
Segregation of roles and permissions
Pre-certified credit bureaus and data sources
Without integrating each provider one by one.
Machine learning without your own infrastructure
Certified security
Infrastructure and scalability
Total cost and time to production (TCO)
AI agent inside the editor
The agent proposes; the team reviews and publishes. Available in binary and calculation nodes.
What can be stated about the engine, and where does each figure come from?
Every claim with its context and its source, so you can check it during due diligence.
- ✓
Policy changes that used to take months are now managed in hours
Self-service for the risk team, within defined permissions and controls. Reported by a customer in production.
Source: Testimonial from Felix Diaz, Credit Risk Director at CredijamarAs of July 2026
- ✓
Automatic versioning with controlled reactivation of previous versions
Every version is recorded and can be reactivated in a controlled way, without an IT deployment.
Source: uFlow security: versioning, roles and access controlAs of July 2026
- ✓
Typical implementation in weeks
Depending on each institution's integrations, security and approval processes. SaaS model with no infrastructure to maintain.
Source: uFlow success story in the cloud: serverless architectureAs of July 2026
- 99,995%
Average uptime (availability)
Committed availability SLA of 99.5%; the platform's observed average uptime comfortably exceeds it (>99.95%).
Source: uFlow availability monitoring (contractual SLA: 99.5%)As of July 2026
- +300 variables
High-complexity analytical models
The engine runs credit models with 300+ variables per decision —such as Lulo Bank's— without degrading response time.
Source: uFlow — event with TransUnion, Bogotá 2024As of September 2024
Why is a modern decision engine imperative today?
Credit-market figures, each with its source. The context that explains the urgency of deciding better and faster.
- 70%
Adults with an account at a financial institution or mobile-money provider in Latin America and the Caribbean
Population ages 15+ in Latin America and the Caribbean (excluding high-income economies). 2024 data published in the Global Findex 2025; the exact indicator value is 69.7%, up from 67.1% in 2021.
Source: Banco Mundial — Global Findex Database 2025 / World Development Indicators, indicador FX.OWN.TOTL.ZSAs of January 2024
- 79%
Adults with an account at a financial institution or mobile-money provider worldwide
World population ages 15+. 2024 data from the Global Findex 2025; the exact indicator value is 78.7%. Serves as a benchmark against the 70% for Latin America and the Caribbean.
Source: Banco Mundial — Global Findex Database 2025 / World Development Indicators, indicador FX.OWN.TOTL.ZSAs of January 2024
- 69,4% vs 49,5%
Approval rate with alternative data versus the same model without it
Measured on applicants with no credit history. 146,036 credit-card applicants in Mexico with no credit-bureau score, served by RappiCard (Banorte + Rappi). The full model, which incorporates delivery-app transaction data, approves 69.4% at a profit-maximizing threshold; the same model without that data approves only 49.5%. Data covering 2021–2024.
Source: NBER Working Paper 33208 — Chioda, Gertler, Higgins y Medina, "FinTech Lending to Borrowers with No Credit History"As of January 2024
- 0,791
Predictive power (AUC) of a machine-learning model using alternative data on applicants with no credit history
Same universe of 146,036 Mexican applicants with no bureau score. The 0.791 AUC exceeds the 0.7 threshold considered desirable in data-rich environments and sits at the upper end of AUCs obtained with alternative data in middle-income countries.
Source: NBER Working Paper 33208 — Chioda, Gertler, Higgins y Medina, "FinTech Lending to Borrowers with No Credit History"As of January 2024
- 30%
Applicants who would be excluded from credit if the decision relied solely on credit-bureau information
Mercado Crédito's portfolio in Argentina, analysed by BIS and FSB researchers using proprietary Mercado Libre data. 30% of the target audience would be rated "high risk" by the local bureau and excluded from the programme; the proprietary scoring model is able to serve that segment.
Source: BIS Working Paper 779 — Frost, Gambacorta, Huang, Shin y Zbinden, "BigTech and the changing structure of financial intermediation"As of January 2019
- 25,4%
Formal firms in Latin America and the Caribbean that are credit constrained
More than 65,000 formal private-sector firms across 109 economies, surveyed by the World Bank (Enterprise Surveys). In Latin America and the Caribbean, 19.4% are partially constrained and 6% fully constrained. The global average is about 30%.
Source: Banco Mundial — Policy Research Working Paper 10502, Islam y Rodriguez Meza, "How Prevalent Are Credit-Constrained Firms in the Formal Private Sector?"As of January 2023
- 27%
Firms that do not apply for credit because terms and conditions discourage them
Same universe of more than 65,000 formal firms across 109 economies. 51.5% do not apply because they already have sufficient funds, 27% self-exclude because of terms and conditions, and only 21.5% actually apply. Of those who apply, 75.8% are fully approved, 9% partially and 8.7% rejected.
Source: Banco Mundial — Policy Research Working Paper 10502, Islam y Rodriguez Meza, "How Prevalent Are Credit-Constrained Firms in the Formal Private Sector?"As of January 2023
- 55%
Potential MSME financing demand left unmet in emerging markets
Formal MSMEs in emerging market and developing economies. Against potential demand of US$10.3 trillion, only US$4.6 trillion was supplied, leaving a US$5.7 trillion gap equal to 19% of those countries' combined GDP. Reference data 2019, published March 2025.
Source: IFC / SME Finance Forum — "MSME Finance Gap: An updated Estimation and Evolution of the MSME Gap in Emerging and Developing Markets"As of January 2025
- 18%
Latin America and the Caribbean's share of the global MSME financing gap
Out of a US$5.7 trillion total gap across emerging and developing markets, Latin America and the Caribbean is the second-largest contributor, driven mainly by Brazil (US$593 billion). Reference data 2019, published March 2025.
Source: IFC / SME Finance Forum — "MSME Finance Gap: An updated Estimation and Evolution of the MSME Gap in Emerging and Developing Markets"As of January 2025
- -16%
Decline in the supply of formal MSME financing in Latin America and the Caribbean
Change in the supply of formal MSME financing in Latin America and the Caribbean between 2015 and 2019. In 2019 such financing averaged 6% of the region's countries' GDP; the contraction was concentrated in three of the largest economies.
Source: IFC / SME Finance Forum — "MSME Finance Gap: An updated Estimation and Evolution of the MSME Gap in Emerging and Developing Markets"As of January 2025
- 20%
Reduction in application processing time at technology-based lenders versus traditional lenders
US mortgage market, loan-level application and origination data from 2010 to 2016. FinTech lenders process applications about 20 percent faster, controlling for loan, borrower and geographic characteristics, with no increase in defaults.
Source: Federal Reserve Bank of New York — Staff Report 836, Fuster, Plosser, Schnabl y Vickery, "The Role of Technology in Mortgage Lending"As of January 2018
- 75%
Financial firms already using artificial intelligence
Joint Bank of England and Financial Conduct Authority survey of UK financial firms, third edition, published November 2024. A further 10% plan to adopt it within three years. In the 2022 edition the figures were 58% and 14%.
Source: Bank of England / FCA — "Artificial intelligence in UK financial services — 2024"As of January 2024
- 2%
Artificial-intelligence use cases with fully autonomous decision-making, without human oversight
Use cases reported by UK financial firms in the 2024 Bank of England and FCA survey. 55% of use cases involve some degree of automated decision-making, but only 2% are fully autonomous: human oversight remains the norm.
Source: Bank of England / FCA — "Artificial intelligence in UK financial services — 2024"As of January 2024
- 24%
Financing applicants that received none of the amount they sought
7,653 responses from small employer firms with 1–499 employees across all 50 US states, surveyed September–November 2024 by the 12 Federal Reserve Banks. 41% received the full amount sought, 36% only part and 24% none.
Source: Federal Reserve Banks — "2025 Report on Employer Firms: Findings from the 2024 Small Business Credit Survey"As of January 2025
- +11% / −20%
Fintech vs. traditional-bank origination
Colombia, May 2023 vs May 2024: fintechs grew new lending 11% while the traditional and cooperative sector fell 20%.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 12% → 30%
Fintech share of unsecured personal loans
Colombia: fintechs went from 12% (2021) to nearly 30% (2024) of unsecured personal-loan origination.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 94%
Fintech concentration in short-term unsecured loans
94% of fintech origination in Colombia is short-term unsecured personal loans.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 34% → 43%
Growth of the subprime segment (2021-2024)
Colombia: the subprime segment rose from 34% (2021) to 43% (2024), driven by macro deterioration and fintechs' inclusion role.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 38%
Consumers whose credit score worsened in a year
Colombia: 38% of consumers saw their credit score deteriorate in the last year.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 60%
Fintech users shared with traditional banks
Nearly 60% of fintech users in Colombia also hold loans in the traditional sector; their debt burden reaches 45% of income.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 20% vs 46%
Roll rate to advanced delinquency in payroll loans (fintech vs traditional)
In payroll-deduction loans, fintechs' roll rates to advanced delinquency (20%) beat the traditional sector (46%).
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
- 39% → 49%
$200k-500k loans in short-term fintech lending
Colombia: loans between COP 200,000 and 500,000 grew from 39% to 49% of short-term fintech volume.
Source: TransUnion — Colombia credit report, Q2 2024 (presented at uFlow event, Bogotá 2024)As of June 2024
How do you give the business autonomy without the bank losing control?
Your risk team changes policies without requiring an IT development for every change. The institution keeps the governance, traceability and security its regulator requires.
Policy governance
Automatic versioning, testing environments and controlled deployment. Change a policy in hours, with controlled reactivation of previous versions.
Traceability and audit
Every decision records its input, the rules applied and the outcome. Explore past transactions: the basis for audit and compliance.
Certified security
ISO/IEC 27001:2022, 2FA authentication, JWT tokens and encrypted data on serverless cloud infrastructure.
uFlow does not replace your IT team: it reduces their involvement in day-to-day policy changes so they can focus on integration, architecture, security and governance.
Everything you need to know
Does uFlow replace my IT team?+
No. It gives the risk and business teams the autonomy to design and change credit policies without writing code, while IT keeps control through roles, permissions, testing environments and the REST API used for integrations.
How does uFlow govern credit policy changes?+
Every policy has automatic versioning: you can store all of its versions, test them in testing environments and reactivate a previous version instantly. Per-user permissions define who can edit, test and deploy to production.
Are decisions auditable for a regulator?+
Yes. Each decision records its input, the rules applied and the outcome, and you can explore previous transactions. That end-to-end decision traceability is the basis for audit and compliance work.
What security certifications does uFlow hold?+
uFlow holds the ISO/IEC 27001:2022 certification, and uses 2FA authentication, JWT tokens and encrypted data on serverless cloud infrastructure.
How does a decision engine help detect over-indebtedness?+
It centralizes queries to bureaus and data sources in a single point, giving a full view of the applicant's debt across the whole system, not just your institution. In Colombia, nearly 60% of fintech users also hold debt with traditional banks; spotting that burden early, before delinquency, is key (source: TransUnion, 2024).
Why are fintechs growing while traditional lending shrinks?+
Because they capture underserved niches —young, new-to-credit, informal-income— with better origination tools. In Colombia, between May 2023 and 2024, fintech origination grew 11% while traditional banking fell 20% (source: TransUnion, 2024). The decision engine is what makes it possible to lend well in those segments.
Why choose uFlow over a large, well-known provider?+
For evidence and closeness, not size. The platform is ISO/IEC 27001:2022 certified and there are verifiable public results: Colsubsidio cut its delinquency from 25% to 12% and Lulo Bank increased processing capacity by nearly 300%. On top of that comes what current users highlight most: a team that supports integration and testing directly, with no layers of support tiers. Every claim we make is published with its source so you can check it during due diligence.
Build, buy, or choose uFlow
We review how you decide today and show you how it would work in the engine, with your own sources and policies.