@to Thanks Trevor!

@harryraymond Hey Harry, Thanks for the question! We already support Messenger and planning to support SMS shortly.

@nitinbajaj1423 Hey Nitin! Thanks for your thoughts here. You definitely raise some interesting points. Would be happy to setup a call and hear any other ideas you have. You can ping me via e-mail here: tom@sagetalk.io. Thanks! Tom

@neobine Hey thanks for bringing that up. The way it works is, if a customer asks a question that the bot doesn't know, the bot will seamlessly transfer the customer to a human. The SageTalk 'Brain' makes it extremely easy for anyone on your team to later jump in and quickly add the correct answer. This way, the next time your customer asks the question, the bot will now be trained with this question and answer it correctly. Check out the link below. https://imgur.com/DJ2xGhi

@patbukala Hi Pat, yes I noticed that it responds on the second try etc but it was that first time that threw it for me... not trying to be a smart ass just that it it broke expectations right away with something that seems easy to solve. It should probably reply like in your pic the first time. Many (most?) will not try a second time and so miss all the good stuff it CAN provide. But enough of the arm-chair quarterbacking best of luck with your endeavor ;D

@neobine Hey there! Right now we are laser-focused on providing the ultimate solution for Chat. Once we feel like we've completed our mission with Chat (automating 100% of queries), then we will probably move on to voice.

@aaron_cohn Great question Aaron, and thank you for raising it. The process for kicking it to a human is part of our patent-pending design. The process is as follows: 1. You ask a question to the bot 2. Bot checks it's brain and sees if it has the answer 3. If it has the answer, it will return to it the user, If it doesn't, it will give you the option to instantly transfer to an agent. The handoff is seamless, and this exact feature is the reason why it made sense to build this product in the first place. We didn't see any product in the space that was doing the hand off right. And we found that with the handoff missing, it was unlikely for any Customer Service teams to adopt the AI into their stack. Turns out we we're right - as today, the only companies that have a working Chatbot + Live Chat solution are big companies like Amazon or Apple, which built their own custom solutions. Hope that helps Aaron, Tom

@tkornblit @aaron_cohn Very interesting product. A common issue with NLP like this (that we're stuggling with ourselves too) is that the bot mis-identifies the intent of the user so even though it should trigger the handoff, it doesn't cause it thinks the user asked a question that matches one of the intents, even though it doesn't. I'm wondering, how do you deal with situations like those?

@aaron_cohn @juhankaarma Thanks for the kind words Juhan! Great question! Our solution to this problem is we built a technical abstraction we call the Brain. The Brain's primary function is to allow a human to train the bot in a very straightforward way, leaving all the complexities to the AI (Intent Matching) Engine, and keeping the user experience very clean. The way you train the bot in our software is as follows: Say you wanted to teach the bot how to answer a question about where your company is located: In SageTalk's Brain, you'd create a topic. As one of the fields in the topic view, you would add the answer to this question, which in this case would be: "We are located in the heart of the wonderful one and only San Francisco." And then you'd add Expressions (meaning different ways you can express the question). In this case expressions could be as follows: 1. where is your head office 2. where is sagetalk hq 3. where is sage ai located 4. where is your main office located 5. where is your company located You'd then save the topic, and then the bot would add all of this information into it's Brain (Massive NLP Search Index). This is also part of our patent-pending design, and what makes our product so unique. Hope that helps, Tom