{"slug": "i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey", "title": "I Built a Voice AI Assistant for Indian Farmers — My 10-Day Voice Agent Journey", "summary": "A developer built Farm & Field, a multilingual voice assistant for Indian farmers, during Murf AI's 10 Days of Voice Agents challenge. The agent supports Malayalam, English, and Manglish, uses tools for real-time market and weather data, and includes features like memory with consent, human escalation, and a specialist agent for crop diseases. Built with Deepgram, Google Gemini, Murf Falcon, and LiveKit, the project aims to make agricultural information accessible via natural conversation.", "body_md": "What if a farmer could simply pick up a phone, speak naturally in Malayalam or English, and ask:\n\n“*What is today's market price?*”\n\nor\n\n“*It's going to rain tomorrow. Should I spray my crops today?*”\n\nor even:\n\n“*My crop is getting damaged. Can someone help me?*”\n\nNo typing. No complicated menus. Just a conversation.\n\nThat was the idea behind Farm & Field, the voice agent I built during the 10 Days of Voice Agents — **VoiceForBharat** Edition challenge by **Murf AI**.\n\nOver these 10 days, I went from building a basic voice conversation to creating an agricultural voice assistant with memory, real-time tools, outbound calling, human escalation, call analytics, multilingual conversations, and a specialist agent.\n\nI built Farm & Field, a voice assistant designed to help farmers with everyday agricultural questions.\n\nFarm & Field can:\n\nThe AI should not try to do everything itself.\n\nIf it needs current information, it uses a tool.\n\nIf the problem needs specialized knowledge, it hands the conversation to a specialist.\n\nIf the situation is beyond what the AI can safely handle, it asks a human for help.\n\nA voice agent is basically four main components connected together:\n\n**Speech-to-Text (STT)** — converts the farmer's speech into text. I used Deepgram.\n\n**LLM** — understands the conversation and decides what to do. I used Google Gemini.\n\n**Text-to-Speech (TTS)** — converts the response back into voice. I used Murf Falcon.\n\n**Real-time transport** — carries the audio between the user and agent. I used LiveKit.\n\nOn top of this, Farm & Field has:\n\n**🇮🇳 Malayalam, English and Manglish support**\n\nFarmers don't always speak in one language.\n\nFor example:\n\n\"*Ente rubber crop-il leaves yellow aakunnundu.*\"\n\nThe agent can recognize Malayalam/Manglish conversations and adapt the TTS voice accordingly.\n\nThis made the conversation feel more natural for Indian users.\n\n**Murf Falcon for fast voice responses**\n\nI used Murf Falcon for TTS because latency is very important in voice conversations.\n\nA slow response makes a voice agent feel like a chatbot that is reading messages aloud.\n\nWith a faster TTS response, the conversation feels much more natural.\n\n**Memory with consent**\n\nFarm & Field can remember useful information about returning farmers, such as:\n\nBut the agent asks for permission before saving information.\n\nThis was important because memory should be useful without automatically storing everything a user says.\n\n**Tools instead of guessing**\n\nThe agent doesn't rely on the LLM to guess live information.\n\nFor example, when a farmer asks:\n\n\"What is today's rubber price?\"\n\nthe agent uses the market-price tool.\n\nThe same approach is used for weather information.\n\nThis makes the system more reliable.\n\n**Human escalation**\n\nThe agent can recognize situations where AI should not be the final answer.\n\nI implemented escalation for cases such as:\n\nThe farmer gives consent before a human-help request is created.\n\n**Specialist agent**\n\nI created a separate Crop Problem Specialist for crop disease and pest-related conversations.\n\nThe main agent can hand the conversation to the specialist without making the farmer repeat everything.\n\nThe specialist also has a different voice, making the handoff clear to the user.\n\n**Outbound calls**\n\nFarm & Field can initiate calls for specific alerts, such as:\n\nI used LiveKit SIP for the outbound calling workflow.\n\n**Call analytics**\n\nThe system also tracks call information such as:\n\nThis helped me understand what happened during conversations instead of treating every call as an isolated interaction.\n\n**Repository**: [https://github.com/safdarsidhik/murf-livekit-starter/tree/main](https://github.com/rohitkumar31/murf-livekit-starter)\n\nBuilt with **Murf Falcon** — the fastest TTS API I used across this build\n\nPart of 10 Days of Voice Agents — **VoiceForBharat Edition by Murf AI**\n\nIf you're building your own voice agent for an underserved use case, happy to answer questions — drop a comment.", "url": "https://wpnews.pro/news/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey", "canonical_source": "https://dev.to/safdar_52ae400ba6ff01703e/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey-l97", "published_at": "2026-08-15 05:59:16+00:00", "updated_at": "2026-08-15 06:11:19.117915+00:00", "lang": "en", "topics": ["artificial-intelligence", "natural-language-processing", "ai-agents", "ai-products", "developer-tools"], "entities": ["Murf AI", "Deepgram", "Google Gemini", "LiveKit", "Farm & Field", "VoiceForBharat"], "alternates": {"html": "https://wpnews.pro/news/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey", "markdown": "https://wpnews.pro/news/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey.md", "text": "https://wpnews.pro/news/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey.txt", "jsonld": "https://wpnews.pro/news/i-built-a-voice-ai-assistant-for-indian-farmers-my-10-day-voice-agent-journey.jsonld"}}