What a wild ride all of this has been. 2 years we've been around now. This all started because I ran an agency after quitting corporate America and needed a white labeled chatbot builder. we've had many up & many downs. The W's keep flying and we persist through the rest because of an amazing community and a platform built for the people using it; by the people using it. 2 years later, our builder has changed a bit and the same faces are still around to see it. We've stepped into a stage where we are outgrowing some of the vendors and platforms we were built upon -- we never want you to experience that so we continue to innovate. Last friday, we had one last problem with a vendor that was a final straw for us coupled with some other things. I'll be releasing version 3 early which is an independence update. We are moving our frontend to our own custom code to match our full enterprise backend to make yours, and your clients, experience much better. Along with that, we will have some independence from 3rd party platforms like OpenAI with optionality to use our in-house LLM specifically trained on casual language and appointment booking (with NLP training for both chat and voice cadences). Couple with this, a couple of other things should come with it: - Flowbuilder production v1.0.0 - conversational pathways with determistic tool calling so you never have to worry about objectives being missed. This introduced global nodes, forced tooling and light weight nodes for a fast and no-hullicnation AI - new front end - we've be oauthing to the new front end so you can acutally choose hwich one you want until legacy gets deprecated - more fallbacks - we'll balance everything all natively and in 3 databases (google cloud, aws, and cloudflare) so we can be 'always up' even if the front end isnt - focus on inhouse telephony solutions - number anayltics, a warming network amoungst users (platform numbers will call platform numbers to keep each other warm) and inhouse verification - all compliance - hipaa gdpr soc2 etcetc - ...tons more

We are refreshing a lot of our content to make sure you have all up-to-date information on using the tools and abilities in the assistant builder :) This one is reading a simple qualification and booking AI with extremely accuracy, making it super scalable to use and never touch again. The prompt is not long, its very short - but you can see how I use the tools to make sure its robust. The biggest thing here? It's not over complicated - it doesn't need to be. This is a super simple AI, the prompt doesn't even scroll down and it just works. You dont need to get fancy in prompting, you can use the tools to do everything you need. We go over: - Prompting practices: how I get my AI to be extremely casual with 1-sentence replies. - Booking: How to make sure you get your booking prompts repeatable and scalable - Identity & Personality Traits: Why they are so important to verbiage and cadence - Reliability: How to get the AI to do what you need it to, every time. - The use of adjectives: Using adjectives affirms HOW you want things done in your task framework. - Testing & refinement: Taking things we don't like and turning them into things we love. - Knowledge base: stop crowding your prompt when you can use RAG. Here is the bot ID you can copy (import bot by bot ID): 1752073640912x549424939141693440 Here is the scrape_website tool ID you can copy (import tool by tool ID): 1725980715689x739519088135241700 Did I leave anything out? Sure. I would test edge cases to make sure it performs under pressure and to give it personality. I would run personality tests against it to see if it reacts the same for different communication types. My build cadence usually goes: - What do I need it to do? I need it to X, Y & Z. Boom theres your task-based framework. - Do I prefer it does it a certain way? Boom, there is your task's adjectives. - How do I want my AI to behave? There is your identity. - Do I want the AI to react differently for certain things? There is your personality traits.