Angel investing in AI
AI is the fastest-growing pitch category in the region right now — and also the easiest one to overstate. The core diligence question is simple to ask and hard to answer: what happens to this company if the underlying foundation models get better?
Why the opportunity is real
Talent arbitrage
A deep, comparatively lower-cost regional engineering talent pool has made "build with LATAM talent, sell globally" a credible model for AI-enabled products, echoing the same pattern already established in SaaS.
Underserved local languages and data
Spanish- and Portuguese-language voice, document and customer-service applications, plus AI applied to region-specific data (agricultural, financial, healthcare records), remain comparatively underserved by global players optimizing primarily for English.
An unsettled regulatory landscape
Unlike the EU’s AI Act, most of the region does not yet have a comprehensive AI-specific law — Brazil has draft legislation moving through Congress — which currently means fewer compliance hurdles but also more regulatory uncertainty than founders may let on.
What to check before investing
Wrapper versus moat
Ask directly whether the product is a thin interface on top of an existing foundation model API (OpenAI, Anthropic, Google) or whether it has a genuine moat — proprietary data, workflow integration, or distribution — that would survive the underlying model becoming a commodity.
Compute cost and burn rate
AI inference and fine-tuning costs can make gross margins look worse than a typical SaaS company at the same revenue stage — get real unit economics per customer or per query, not just top-line growth.
Data rights and provenance
Confirm the company has clear legal rights to the data it trains on or processes, particularly for anything touching regulated data (health, financial, biometric) — this is a fast-evolving legal area and a real diligence gap in many early pitches.