A Filipino Startup Built AI That Actually Knows Philippine Law — 95% Accuracy on the Bar

The problem nobody wanted to solve

Ask ChatGPT about Philippine labor law and you'll get a confident-sounding answer with a 40% chance of being wrong. Not because the model is bad — because the data it was trained on is broken. Philippine legal information is technically public, but scattered across more than 150 national government agencies and departments, each maintaining their own website at varying degrees of reliability and completeness. Many don't have historical issuances online at all.

That's the insight behind Anycase.ai — a Filipino legal technology startup that presented at OpenAI's Founder Day in Singapore last week, one of only five Southeast Asian startups invited to demonstrate their products. CEO Beato Bongco's message was blunt: the real problem isn't AI capability. It's the data.

What they built — and how hard it was

Anycase built what they call a Legal Intelligence Layer: a proprietary database of Philippine legal materials assembled not from web crawling, but from physically going to agencies and asking for the records.

The process: request documents from agencies, digitize printed materials, apply OCR, clean the data, and have lawyers and paralegals verify every document before it enters the platform. One employee's full-time job is visiting government agencies to collect legal records. That process can take two weeks to three months per source.

"We know how this sounds. It's slow. It's expensive. It doesn't scale the way software is supposed to scale. But here's the thing: no one else was doing it. And without this data, everything else we build is just a better search box on top of the same broken foundation." — Beato Bongco, CEO, Anycase.ai

On top of that database, Anycase built proprietary retrieval and reranking models, plus evaluation benchmarks built from Philippine Bar Examination questions and specialized practice areas like banking and finance. The result: 95% accuracy on Bar exam-style evaluations — a benchmark that requires understanding specific Philippine jurisprudence, not just general legal knowledge.

The numbers in its first year

• 8,000+ users
• 20x revenue growth
• Cash-flow positive
• 75% reduction in legal research time reported by users

During the OpenAI Founder Day demonstration, Bongco showed the platform analyzing a labor dispute in real time: identifying legal issues, retrieving relevant statutes, regulations, and Supreme Court decisions, outlining available remedies, and helping prepare procedural requirements for cases before the National Labor Relations Commission.

Why non-lawyers should pay attention

Anycase is built for lawyers, but the problem it's solving matters to anyone running a business in the Philippines. Philippine law is what governs your employment contracts, your lease, your online selling terms, your BIR obligations, and the compliance requirements for whatever industry you're in. Most small business owners handle these with a mix of Google searches, informal advice, and hoping for the best.

Tools like Anycase — which understand Philippine-specific law with documented accuracy — change that calculus. A solopreneur checking whether their freelance contract is compliant, an online seller wanting to understand their consumer protection obligations, or an SMB owner handling an NLRC complaint: all of these are cases where accurate, locally-grounded legal AI matters. Right now Anycase is aimed at legal professionals, but the direction is clear.

The broader signal

Co-founder Gio Tiongson summed up the mission at the Singapore event: "I've always believed the Philippines should not just be a market that imports software built elsewhere. We can build deep, local, domain-specific AI too."

That's the bet. The Philippines has domain knowledge — about its own laws, its own tax system, its own market — that no US or Chinese model is going to replicate from web data alone. Anycase is one of a growing group of Filipino startups (alongside NMBLR.ai, Salig AI, DocTax) betting that locally-built AI wins on local problems. In Anycase's case, the first year of results suggests the bet is paying off.

Source: Pinoy startup says fragmented legal data holds back AI for PH law — Newsbytes.PH