Vintage 1, year 2 of investing — first cohort live.
A repeatable loop. Six stages. 3–4 ventures a year.
A repeatable loop to launch 3–4 ventures per year and compound the winners for decades.
Research the wedge
find workflows where AI creates 10× advantage.
Partner with operators
domain experts + builders.
Build fast
small team, AI-powered core, rapid iteration.
Earn traction
real users, real revenue, honest signals.
Prove unit economics
LTV:CAC matters from day one.
Compound
scale profitably, build durable moats.
Jess and Andrea spent the prior decade building inDinero ($100M+ ARR, 100+ headcount) and investing out of San Francisco. Amanda and Felipe earned operational scar tissue in Brazil — Amanda scaling Sigga through a 10× exit, Felipe building Bamboo DCM in the capital markets. We met in 2024 when all four of us arrived at the same conclusion from opposite angles: Brazil has massive service-economy volume, underestimated senior engineering talent, and almost no pre-traction capital with operational depth attached. The division of labor is deliberate: the São Paulo team operates; the San Francisco team brings playbook + capital + network.
Three of the ten rules we operate by.
Not values. Rules that constrain behavior — the decisions we have already made so we can spend operating time on the ones that are left.
First ticket or no ticket.
We are first-money-in or we do not invest. The studio's edge is at the founding moment.
3–4 ventures per year. No exceptions.
We deliberately cap throughput. Studios that scale faster than their stack matures debase the very advantage that justifies the model.
What we will not do is what we are.
A studio that cannot articulate its negative space cannot articulate its positive. Saying no is the discipline.
Let's talk.
Whether you're a founder building or an investor looking for AI-native exposure in Brazil — we'd love to hear from you.