Kaagaz_India
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Kaagaz checks a family’s welfare documents before they apply, by camera and voice, in their own language. Families in India lose benefits over small paperwork problems: a name spelled two ways, a date of birth that doesn’t match, an income certificate that’s too old. They usually find out weeks later, after a wasted trip to the office. With Kaagaz, a family films their papers or adds photos. SAM 3.1 finds each document, Muse Spark reads it, and plain rules check every family member against five Maharashtra schemes. An agent builds a fix plan from verified procedures only and reads it aloud in Marathi, Hindi, English, Tamil, Telugu or Kannada. Families can ask follow-up questions by voice, and code fact-checks every answer before it’s shown. The AI reads and explains; it never decides eligibility. The demo uses synthetic SPECIMEN documents only.