This may be more of a GHL question, but for some reason, LeadConnector has started to automatically send this text to callers who book through my assistables.ai voice agent: "Your calling preferences have been updated and your number has been added to an internal DNC list. Reply STOP to unsubscribe." It didn't do this a few days ago. No change to the workflow the agent is using. No DND fields are checked or anything like that. Nothing in the audit logs. When I click the three dot details in the conversation stream I see the attached. Anyone know how to avoid this (clearly annoying and not needed) SMS to users?

Hey Assistable community! 👋 I’m building out voice AI assistants for clients (inbound/outbound calls via dedicated phone numbers), and I have a question about how the AI learns/adapts from real conversations. Scenario: My client calls the AI’s designated phone number and shares specific info (e.g., details about their business, preferences, products, FAQs, etc.) – essentially “training” it verbally during the call. 1. Will the AI store and remember everything shared in that conversation for future interactions (with that contact or generally)? 2. More importantly, can the client effectively “train” the assistant this way – meaning, does the info get pulled into the permanent conversation memory, or even better, auto-indexed/used to improve the knowledge base over time? I know Assistable has permanent conversation memory and KB features (with thumbs up/down voting in the inbox), but I’m curious if verbal/phone-based sharing translates directly into persistent learning for the assistant. Has anyone set this up for clients who want to self-train their AI without going into the dashboard? Any tips or gotchas? Thanks in advance – love how powerful this platform is! 🚀

Been struggling lately with one of my global prompts giving me consistent results. Won't go into all the details but having issues like repeating itself, saying what it is doing out loud vs behind the scenes, not hanging up after call, etc. Plenty of prompt troubleshooting and editing hasn't been a home run. I had been using GPT 4.0 forever and experimented with 5.0 today. Haven't had a chance to look at the others yet. 5.0 is not much better, really. **Any recommendations for the best model these days?**

Anybody else having issues with the assistant checking availability? It never seems to go through. Just stalls.

Hi everyone, I just read a LinkedIn post where someone lost $20K in minutes because their Voice AI callback form was hit with fake international numbers. Their agent kept calling and talking, draining their balance completely. This is a wake-up call for us. We should make sure we’re fully covered against something like this — things like CAPTCHA, spend caps, geo locks, concurrency limits, etc. I wanted to bring this up here so we can make sure we’re all covered and avoid any risks like that and protect ourselves and our clients. Better safe than sorry. 🙃 We’ve seen a few of our clients go through this too, and it can get really messy really fast. That’s why I think it’s important we put the right safeguards in place together. 🙌 @Assistable Team @Mike Copeland @Jorden Williams