We tested whether Claude for Financial Services works in LATAM
Spoiler: almost everything that matters is already available. We put Claude for Financial Services to the test with public APIs from Brazil, Peru, and Chile.
It's 10 p.m. at a credit rating agency in Lima. A junior analyst is 11 hours into assembling the first draft of a rating surveillance review. She has 14 browser tabs open. Committee is in 36 hours. This scene is a common one at many rating agencies across Latin America.
When Anthropic launched Claude for Financial Services, the obvious question for us wasn't whether it worked on Wall Street. It was: does it work in LATAM? We tested it with four public APIs (BCB, Receita Federal, OFAC, CVM) and Claude Sonnet 4.6 orchestrating via MCP. Target: Grendene S.A., a B3-listed Brazilian issuer. In 4 minutes and 12 seconds the system produced the foundation of a credit memo that, in our experience, often takes two eight-hour shifts.
What we found: the infrastructure already exists. The APIs are available today, for free. Claude already reasons in financial Spanish and Portuguese. The commercial moat isn't the data, it's what sits on top: multi-jurisdictional schema, regional prompt packs, compliance overlay. The window to assemble this closes this year, before Globant, Softtek, or the regional Big 4 claim the space.
Read the full post on MediumRequest demo access
If your team lives this workflow, we'll put together a 30-minute diagnostic on which part of your process can be automated today with public APIs + Claude. No commitment, no corporate deck: a technical conversation.
Contact: luis@bankinglatam.com