Defined and built
Analysis flow, documents by applicant type, outputs and controls.
BankIA / Platforms / DatIA Recaudo
DatIARecaudo
Under construction · for banks, cooperatives and natillerasCredit analyses in hours, with every figure backed by its document.
DatIA Recaudo reads the applicant’s documents, organizes the information, calculates repayment capacity using verifiable rules and delivers an analysis with risks and a recommendation for the committee. AI prepares; the decision always rests with a person.
- AI does not make up figures
- Evidence for every data point
- Human decision
- Built for Colombia
Solicitante · IndividualCertified income · 4 documents read
CompleteEstimated repayment capacityCalculated in code · see formula
$1,480,000AlertEmployment certificate dated 5 months ago
ReviewIllustrative example with fictitious data.
AI · example
Suggestion for the analyst: request a recent employment certificate. Source: page 1 of the uploaded certificate. Medium confidence.
Evidence in the documentA person decides
What it is, what it is not yet
We are building DatIA Recaudo with the rules of the BankIA platforms. Before opening it to the public we will run tests with fictitious data and an independent security review.
First version in internal testing
Encrypted documents, repayment capacity and score calculation, the report and human decision already work with fictitious data. AI document reading is built; it still has to be activated and its accuracy measured.
Before launch
Testing with fictitious data, a security review and a legal review of data processing.
This platform is not available yet. Nothing described here is currently offered as a service in operation.
From document to analysis, in six steps
Every step leaves a record and can be reviewed.
- 1
Application
The institution creates the analysis and chooses the applicant type: individual or legal entity.
- 2
Document upload
The applicant or the advisor uploads the documents through a secure link.
- 3
Reading and verification
The AI extracts the data from each document and tags it with its source page and confidence level.
- 4
Calculation
Code calculates income, expenses, debt level and repayment capacity with visible formulas.
- 5
Analysis
A report is assembled with findings, alerts, missing documents and a recommended amount and term.
- 6
Decision
The committee or analyst approves, adjusts or rejects. A record is kept of who decided, when and why.
What it reads depending on who applies
Each institution defines the list it requires. The platform uses it to detect what is missing.
| Applicant type | Usual documents |
|---|---|
| Individual empleada | Cédula, employment certificate, pay stubs, bank statements, income tax return if applicable. |
| Individual independiente | Cédula (ID card), RUT, bank statements, income tax return (declaración de renta) or an income certification from an accountant. |
| legal entity | Chamber of commerce certificate, RUT, financial statements, income tax return (declaración de renta), bank statements, ID card (cédula) of the legal representative. |
The final list is refined with the institutions that test the platform.
An analysis the committee can audit
Repayment capacity
Income, expenses and debt level with the formula in plain view, so anyone can verify it.
Alerts and risks
Expired documents, inconsistencies between supporting documents and missing data.
Suggested amount and term
An explainable recommendation, never an approval.
Complete file
Documents, extracted data, calculations and decision in a single file.
Evidence per data point
Each extracted figure shows the document and page it came from.
History
What was uploaded, what changed and who decided, with dates.
How we keep AI from getting it wrong with your money
In lending, a misread figure is costly. These rules are part of the design from the start.
AI reads; code calculates
Models extract data. Code does the math, using exact decimal arithmetic.
Human review of anything doubtful
Any data point with low confidence is flagged for an analyst to confirm.
Institution rules
Each institution configures its own credit policies; the AI does not change them.
Final human decision
No loan is approved or rejected automatically.
Designed with the Colombian framework in mind
We do not claim compliance until the legal review is done. These are the frameworks that guide the design.
Ley 1581 de 2012
Personal data protection: purpose, authorization, access and deletion.
Ley 1266 de 2008
Financial habeas data: credit information and data subject rights.
Ley 2157 de 2021
Rules on the reporting and expiry of negative data in credit history.
This list is informational and is not legal advice. Each institution is responsible for its own compliance.
Built for those who analyze credit every day
Banks and fintechs
Speed up the analysis of applications with an organized file.
Cooperatives and funds
Consistent analyses for the credit committee.
Large natilleras
A simple analysis before lending to someone outside the group.
Be among the first to try it
Leave us your details and we will write to you when we open testing with institutions.
Frequently asked questions about DatIA Recaudo
Is it available yet?
No. It is under construction. You can join the waitlist.
Does AI approve or reject loans?
No. It prepares the analysis and recommends; a person decides.
What happens if the AI misreads a document?
Each data point carries its evidence and confidence level; anything doubtful is flagged for review, and calculations are done by code.
Where are the applicant’s documents stored?
In the institution's own space, separate from the others. They are kept according to the policy each institution sets and are deleted automatically when it expires; credit bureau reports, at the end of their term.
Does it connect to credit bureaus?
Yes, under the contract the institution has with each credit bureau and with the data subject's authorization. Until there is a contract, it works in demo mode.
Guides on credit analysis
What a credit analysis is and what it checks
What a lender analyzes before approving a loan: income, debts, repayment capacity, credit history and backing, with a worked example and the documents.
Read the guide →How a loan installment is calculated (fixed installment)
A loan’s fixed installment step by step: the formula, how much is interest and how much principal, and how the installment and total cost change with the term.
Read the guide →Help us build the credit analysis you would want to use.
Tell us how you analyze credit today and we will let you know when we open testing.