My client is waiting for me to deliver their bot, that they paid a month ago. They want to call up fresh leads with voice AI. They have a medical knowledge base of 300 rows of data. Since KB for voice doesn't work, I deliberately tested Buildship GSheets & GDocs and even make.com connections. On all my tests the chat bot read all of the data and voice bot only read first 20-30 rows, and said it didn't have access to rest. What else can I do to get this fixed?

Tons of new features, hyper-customization of the voices, and HIPAA compliant calling. Latency is now instant - we will enable inbound, outbound, iframe-initiated + web calling (to have the AI talk in the browser). Switching to 11labs for voices to for broader options and lower latency. Buy twilio numbers through us to link to your assistants for inbound calls to your business. Actions - voice assistants will share the same functionality as the chat assistants. BONUS: I will be putting buttons throughout the platform to walk you/your clients through deployments by talking to them in the browser about how to do everything.

I just finished Jordan’s video on Building a Production AI Voice Bot and had 2 clarifying questions: 1. Does the built-in Assistable knowledge base work with phone-based AI assistant conversations, or is it limited to chatbots and web call bots? 2. Why do we need to build a vector database using Supabase with Proship integration for the knowledge base? Would the built-in knowledge base function the same way, or is there a meaningful difference in capability or use case? Thanks in advance for the clarification!

Great to hear about voice V3. Wouldit reliably find/merge an existing GHL contact and reliably book an appointment used the merged contact in a web call (via programmable button or Orb)? We have been experiencing a persistent booking issue with Voice V2 in that scenario - it cannot find existing contact when given a phone num & it cannot book appt.

Has anyone successfully used the Assistable AI Assistant to book appointments with different durations based on appointment type? If so, how are you handling it? The standard book_appointment function doesn’t seem to support variable lengths.