01
Banking
Credit risk · IFRS 9 · Collections and fraud
III. Sectors · 01
01
Credit risk · IFRS 9 · Collections and fraud
Support activities
Primary activities
In banking, every decision balances growth, prudence, and capital preservation. Safeguarding assets, financing projects, and sustaining long-term relationships require financial strength and sound judgment across retail and wholesale banking.
From funding to recovery, artificial intelligence delivers value across five areas:
Technology evolves. The foundations of trust endure.

Banco BV
2015–2017
ChallengeCapital leaked through how each exposure was classified: incomplete records, collateral missing from the calculation and no lineage, while credit was priced without the capital it consumed.
SolutionA certified exposure base with lineage, a rules and quality engine, a capital simulator by exposure, client and portfolio, and RAROC at the credit desk when the deal is proposed.

Banco BV
2015–2017
ChallengeSeven disconnected routines calculated market-risk capital on aggregated positions and conservative assumptions, raising capital costs and making the stress tests regulators now required slow and unreliable.
SolutionA single positions base with lineage, all seven components in one calculation reconciled with accounting, pre-trade capital, stress scenarios and an architecture ready for sensitivities.

Banco BV
2019–2020
ChallengeAn annual sample-based audit covered a few hundred exposures in portfolios of millions, with weeks of manual reprocessing, while new rules raised the cost of any capital error.
SolutionContinuous control over the whole population at every reporting date: independent recalculation, completeness tests, anomaly detection, automated reconciliations and an evidence trail for the board, auditors and the Central Bank.

Arab Banking Corporation
2018–2019
ChallengeWholesale credit analysis was an annual, hand-built review of the last balance sheet, so the bank spotted a client's deterioration late, losing more and provisioning wrongly.
SolutionAutomated spreading of financial statements with human review by exception, trend and peer analysis, explainable early warning, a portfolio view and a documented signal for provisioning.

Banco Itaú
2013–2014
ChallengeProvisioning, collections and pricing ran on disconnected risk truths: spreadsheets with manual overlays, collections only after delinquency and product price tables, causing avoidable losses and mispriced margin.
SolutionA single PD, LGD and EAD engine, with independent validation and an audit trail, feeding expected loss, preventive collections, limit management and RAROC-based pricing.

Casas Bahia
2020–2021
ChallengeStore credit rates and installments had to reflect risk, term, product and customer profile.
SolutionStudy of risk-based pricing, with a term and installment model for store credit.

Casas Bahia
2020–2021
ChallengeForged documents, synthetic identities and misuse of taxpayer IDs (CPF) threatened credit origination.
SolutionStudy of an identity-fraud score cross-checking CPF, phone, address, device, banQi account and purchases, combined with facial biometrics and liveness detection.

Casas Bahia
2020–2021
ChallengeThe group's integrated customer base had to be activated and engaged in the digital account.
SolutionStudy of segmentation, propensity, acquisition and activation, with next-best-action over the integrated base.

Banco Caterpillar
Grupo Boticário · GAVB · 2021
ChallengeCredit rate offers had to be personalized to each customer's risk profile.
SolutionProposal for reapplying the customer segmentation and collaborative-filtering rate recommendation built in the Creditas MVP.

Creditas
Grupo Boticário · GAVB · 2021
ChallengeVehicle-backed credit rates had to reflect each customer's risk profile and the lender's risk appetite.
SolutionMVP that groups customers into seven profiles and recommends rate bands, restrictive or promotional, through collaborative filtering among customers with similar risk.

Havan
Grupo Boticário · GAVB · 2021
ChallengeCredit origination had to grow with delinquency under control, alongside debt recovery.
SolutionProposal for application and behavioral scoring accelerators and three-tier collections, under three engagement models: ML Discover, an analytics squad or outsourcing.

Sicredi
Grupo Boticário · GAVB · 2021
ChallengeInterest rates had to reflect cooperative members' risk profiles and the institution's risk appetite.
SolutionProposal for applying the Creditas MVP approach: profile segmentation and rate recommendation through collaborative filtering among similar risk profiles.

Banco Digimais
Grupo Boticário · GAVB · 2023
ChallengeAccounting data for several products was processed by hand, making the close slow and error-prone.
SolutionProposal for mapping sources, restructuring the accounting system and building an accounting base by contract, client, product and vintage, with dashboards and RPA for reconciliation and close, over six months.

Financial institution
Grupo Boticário · GAVB · 2023
ChallengeIn a network of thousands of branches, it was unclear what drives efficiency or how to simulate new branches in any municipality.
SolutionA predictive model of the efficiency index, a feature optimizer under business rules, branch clustering and simulators for new and existing units, running monthly.