Pramanik

Pramanik - An open AI accountability office powered by autonomous voice agents.

• Complaints don’t fix themselves. Citizens report civic problems, but chasing officers and checking the outcome still takes time and effort. Pramanik automates that follow-through.
• One issue, shared context. AI agents aggregate related complaints and bring resident reports, photos, previous conversations, and officer commitments into a shared case dossier.
• Coordinated AI teamwork. Built with the OpenAI Agents SDK and GPT-5.4 mini, specialist agents review reports, history, and responsibility. A coordinator turns their findings into an action plan.
• Voice agents that ask for action. A GPT-Realtime-2.1 voice agent, connected through Bolna, calls the designated contact, asks informed questions, revisits previous promises, and requests clear deadlines.
• Follow-ups that continue after the call. Through tool calling and persistent background jobs, agents save commitments, update case records, and schedule the next follow-up.
• “Done” needs a ground check. Completion claims open a public survey where residents submit observations and photos. GPT-powered visual assessment supports human review, and conflicting evidence keeps the case open.
• Accountability everyone can inspect. A public Kanban board, case dossier, and activity summary show published conversation records, promises, evidence, completed actions, and next steps.
• Less chasing, more community participation. Residents can add issues, contribute evidence, and check progress together—making public-service commitments easier to question and verify.