03
Payments
Real-time fraud prevention · Risk scoring at digital payments scale
III. Sectors · 03
03
Real-time fraud prevention · Risk scoring at digital payments scale
Support activities
Primary activities
In the payments industry, every transaction is a decision made in milliseconds: approve or decline, protect without creating friction, and drive growth amid increasingly compressed margins. With instant payments, new market entrants, and AI agents beginning to make purchases on behalf of consumers, the ecosystem is evolving faster than ever.
From acceptance to settlement, artificial intelligence drives value across five key areas:
In an industry where the difference is measured in fractions of a second and percentage points, intelligence is no longer a competitive advantage. It becomes infrastructure.

Casas Bahia
2020–2021
ChallengeApproved sales later turned into disputes, and boleto payments with anomalous behavior also had to be caught early.
SolutionA chargeback probability model to anticipate sales that will be disputed, boleto anomaly detection, delivery verification, dispute handling and customer communication.

Casas Bahia
2020–2021
ChallengeThe card and benefits partnership had to be structured and built into e-commerce.
SolutionSelection of the card partner and benefits, partnership terms and policies, integration of the offer into e-commerce and customer communication.

Callink / Mercado Pago
Grupo Boticário · GAVB · 2021
ChallengeActivation of card terminals among Mercado Pago customers contacted by Callink had to be maximized.
SolutionProposal for propensity models for effective contact, reaching the right person and getting a promise, with a funnel dashboard, deployment and backtesting.

Edenred
Grupo Boticário · GAVB · 2021
ChallengeFleet customers had to be steered to other stations matching their preferences, price per liter and negotiated rate, to strengthen bargaining power with merchants.
SolutionProposal for a two-track recommender: collaborative filtering for known profiles and content-based recommendation for new customers, with backtesting, in three months plus two of assisted production.

Grupo Boticário · GAVB
2021
ChallengeChargebacks caused losses, higher acquirer fees and provisions for future debits.
SolutionBefore the sale, a real-time score of the chance a transaction becomes a chargeback within four months; after it, RPA submits the dispute file to acquirers.

Mooz
Grupo Boticário · 2021
ChallengeThe business needed AI products for credit, collections, marketplace fraud and forecasting of asset demand and prices.
SolutionProposal for combining the credit scoring and collections accelerators, an anti-fraud platform yet to be built and forecasting of asset demand and prices.

Getnet
Grupo Boticário · GAVB · 2022
ChallengeVolume and margin had to be balanced across a large number of products and consumers.
SolutionAI-driven dynamic pricing that recommends the price with the highest conversion probability and the best balance between volume and margin.

Getnet
Grupo Boticário · GAVB · 2022
ChallengeMedia, CRM, monetization and sales teams needed segmentation, personalized offers and pricing, and better offer timing.
SolutionProposal for a data strategy covering vision, culture, governance and self-service, plus use cases for retention, through churn probability, and AML, flagging suspicious customers for review.

Edenred
Grupo Boticário · GAVB · 2023
ChallengeFraud analysis of users, merchants and client companies was manual across the four benefit card products.
SolutionModels per offender: transactional and geographic user anomalies, a merchant anomaly score and new-contract risk, with business rules; samples analyzed fell 70% and frauds detected rose 25%.