Stablecoins, AI Agents, and The Future of Global Banking

What It Covers

Jeeves CEO Dilyp Tasman explains how his company built a stablecoin-native financial operating system across 25 countries, growing TPV from $400M to $3B in two years by owning local infrastructure, securing regulatory licenses, and deploying AI to run operations with 140 people instead of 200.

Key Questions Answered

• •Infrastructure ownership as margin driver: Building proprietary ledger infrastructure across 25 countries — including local card issuing BINs directly with Mastercard — expanded Jeeves' gross margin from 40% to over 80%. Owning the full stack eliminates intermediary costs and enables consistent product experience across Brazil, Mexico, and Colombia without country-by-country rebuilds.
• •Stablecoin country expansion economics: Launching a new country previously required 8 months and $500K–$700K in setup costs. Using USDC as settlement infrastructure, Jeeves launched Argentina in a fraction of that time and cost. The model requires only 2–3 local salespeople to profitably serve mid-market clients, making previously unviable markets like Peru now economically feasible.
• •Enterprise trust over technology pitch: When selling stablecoin-powered payments to CFOs, Jeeves avoids mentioning stablecoin entirely. The product is branded "Jeeves Instant Pay," competing on speed — one hour versus one to two days — and reliability. CFOs care about funds arriving on time, not underlying rails, so brand trust converts faster than technical explanation.
• •AI-driven underwriting leverage: A four-person underwriting team now processes $2–3B in annual TPV using self-learning models, a task that required 15 people two years ago. Jeeves reduced total headcount from 200 to 140 while growing revenue 10x and volume 8x. Document ingestion, KYB, and multilingual customer service all run on internally trained AI models.
• •Segment focus over growth breadth: In 2023, Jeeves deliberately offboarded small businesses and concentrated entirely on mid-market and enterprise clients with revenues between 10M–100M Brazilian reais. This forced the addition of a payments product — accounts payable volume exceeds corporate card volume — and drove the cross-sell model that now generates compounding gross profit per customer.