Legit Ba 'To

Misinformation spreads fastest on Facebook and Instagram, yet most fact-checkers live on English-language websites nobody opens mid-argument. The people most exposed, often older and non-technical Filipinos, have no fast, trusted way to check a claim in their own language, where they actually encounter it.

Legit Ba 'To meets people inside the apps they already use. A multi-agent pipeline scrapes the linked article, searches the open web for corroboration, and returns a plain-language verdict with its reasoning and sources, in Tagalog or English. The same intelligence powers a web chat, an Instagram DM bot, and a Messenger bot from one shared backend.

• —Verdicts across five classes (legit, false, misleading, unverified, or satire), each with red flags, green flags, and cited sources
• —A two-agent research pipeline: a conversational core agent delegates to a research agent that scrapes and cross-checks the web
• —Vision analysis that reads text from screenshots and flags AI-generated images
• —One shared backend serving three channels: web chat, Instagram DM, and Facebook Messenger

“Fact-checking only helps if it reaches people where they already argue, so it lives in their DMs, in their language.”

Ships as a working product that takes a link or a screenshot and returns a sourced verdict in seconds, across web and social DMs. Async message batching, per-sender rate limits, and provider circuit breakers keep the bots responsive under real load, while an admin dashboard tracks live API spend.