The problem
5 billion people worldwide lack affordable legal access. In Latin America, a legal consultation costs what many earn in a week. A single mother in Chile loses her home to a fraudulent auction. A worker in Peru misses the deadline to file a claim. A family in Bolivia doesn't even know they've been sued. This is the universal problem we are solving.
The inspiration
I'm a former mining engineer. I watched automation transform a massive industry in a decade. The natural question was: why does the judicial system still operate like 100 years ago across the entire world, when the technology exists to democratize it globally?
What we built
Legal Opportunity is an AI-native legal operating system that runs an entire law firm with AI. Not a legal chatbot. It's infrastructure: a 17+ step pipeline that executes case classification, precedent research, strategic analysis, document drafting, 24/7 judicial monitoring, and judicial auction scoring.
AI makes critical decisions in production. Humans only supervise and approve final outputs. We start operating in Chile, scale across Latin America, and the architecture enables global expansion.
Key innovation: dual anti-hallucination architecture
In a domain where a legal error can ruin a family, we can't afford "the model sometimes hallucinates." We built a two-layer barrier:
- JSON whitelist of validated legal articles
- Google Search Grounding against official legal libraries
Every legal citation passes through both verifications before reaching the client. This turns Gemini into a trustworthy source for high-stakes legal work, not just an assistant.
Dual architecture: auto + manual (unique global reach)
We designed the system to operate in any jurisdiction worldwide. In countries with digital court portals (Chile, Peru, Costa Rica, Spain, India, Australia, South Africa, Kenya, etc.) we use automated scraping. In countries without digitalization (Bolivia, Guatemala, much of Africa, Vietnam, Cambodia) the lawyer uploads documents to a digital "intake" and AI processes everything.
This means we can operate in ~90 countries from day one, including markets that NO AI legaltech competitor can serve. While EvenUp, LegalMation, and Eudia only operate in digitalized US and European markets, we reach where no one else does.
What we learned
Legal AI shouldn't be a workflow in itself, but part of a human workflow. The lawyer supervises, AI operates. This matches what Mayada Gonimah (Thread AI) recommends in Google Cloud's Future of AI 2026 report.
We also learned that "proprietary data" isn't the laws (which are public), but the accumulated results: what strategy worked for what case type, with what court, against what type of opponent. Every case handled builds an advantage no competitor can replicate.
The challenges
Anti-hallucination in a zero-tolerance domain: months iterating the dual barrier until reaching zero fabricated citations.
Cost/quality balance: we use Gemini Flash for high volume and Gemini Pro only for critical decisions. Per-document cost <$0.06 USD.
Customer acquisition with low digital trust in LatAm. We solved it with freemium: free AI consultations that convert to paid services.
Team: as a non-lawyer founder, finding a lawyer co-founder who understands startups and doesn't ask for 30% equity just for signing.
Global vision
Chile as operating base, Latin America as natural expansion, the world as horizon. Access to justice is a universal problem: 5 billion people without affordable legal access. Our dual architecture lets us serve markets no competitor can touch. This isn't a Latin American legaltech, it's a global platform with Latin American roots.
Log in or sign up for Devpost to join the conversation.