Built a Next.js and TypeScript prototype that combines form-image analysis, document-field extraction, voice responses and review steps to help users work through complex forms. The workflow produces PDF output and supports configurable AI integrations: local Gemma inference through Ollama, a Python-server path and Gemini fallback. Bengali and Hindi interactions are central to the guided experience.
Challenges
- Making complex forms easier to complete through Bengali and Hindi interactions while extracting information from forms and supporting documents.
Solutions
- Connected form-image analysis and document extraction with guided voice responses, review steps and PDF output, using configurable local and cloud model integrations.