Vapi Data Capture + STT Recommendations?
Hey everyone! ๐ I'm building a mental health phone agent with Vapi that needs to collect specific data points (name, DOB, responses to 12 screening questions, notes) and send them to AWS DynamoDB. What's the best approach to capture this structured data from the conversation? Should I: 1. Set up explicit variables in the Vapi agent for each data point? 2. Parse the conversation transcript after the call ends? 3. Use a combination of both approaches? I know Vapi can send webhooks to AWS, but I'm trying to figure out the cleanest way to extract the specific data points from the conversation flow. The agent follows a specific script with clear questions, but I want to make sure I'm reliably capturing the responses. Also, which transcriber (STT) do you recommend using on Vapi between OpenAI Whisper, ElevenLabs, or Deepgram? I need accurate transcription for mental health conversations and accurate patient responses. Has anyone implemented something similar or have recommendations on the best practices? Thanks in advance!