Voice AI Call Router
Budget: ₹1,500 – ₹12,500 INR
I need a working Voice-AI agent that can pick up every incoming call, speak naturally with the caller, and then route the call to the right person or department. The agent’s top priority is real-time call handling; note-taking and knowledge-base lookup can wait for a later phase once the core hand-off works smoothly.
Here’s what I’d like delivered:
• A small, self-hosted or cloud-hosted prototype that answers calls through a service such as Twilio, Asterisk, or a solution you recommend.
• Automatic speech-to-text, intent detection, and a clear routing decision. For now I’m open on whether the logic is keyword-based, rule-based, or uses an LLM—choose the fastest path to a reliable first version.
• Immediate call transfer to the correct phone number or SIP extension. If no route criteria match, the agent should drop to voicemail.
• A lightweight dashboard or log file where I can see call transcripts, routing decisions, and call duration.
Keep the setup simple so I can test quickly, yet structure the code so additional skills (detailed message capture, quick knowledge lookups) can slot in later without rewriting. The solution must run on Linux or a low-cost cloud instance, and all configuration instructions should be clear enough for me to replicate the environment from scratch.
If you have prior work with Twilio Voice, Dialogflow, Whisper, or similar tech, point me to it—proven audio pipeline experience will help us move faster.
Here’s what I’d like delivered:
• A small, self-hosted or cloud-hosted prototype that answers calls through a service such as Twilio, Asterisk, or a solution you recommend.
• Automatic speech-to-text, intent detection, and a clear routing decision. For now I’m open on whether the logic is keyword-based, rule-based, or uses an LLM—choose the fastest path to a reliable first version.
• Immediate call transfer to the correct phone number or SIP extension. If no route criteria match, the agent should drop to voicemail.
• A lightweight dashboard or log file where I can see call transcripts, routing decisions, and call duration.
Keep the setup simple so I can test quickly, yet structure the code so additional skills (detailed message capture, quick knowledge lookups) can slot in later without rewriting. The solution must run on Linux or a low-cost cloud instance, and all configuration instructions should be clear enough for me to replicate the environment from scratch.
If you have prior work with Twilio Voice, Dialogflow, Whisper, or similar tech, point me to it—proven audio pipeline experience will help us move faster.