Please can you help? I have an issue with intents for service complaints. After triggered, my tests shows that customers complaints explantion includes phrases that re-triggers the intent. This causes it to loop back to the beginning. Do you have a solution for this?
Hi, I'm new here and trying to learn how to develop bots with the AI Intent feature. I followed the tutorial step by step but I'm encountering the following problem: COGNITIVE - AI classification failed. No function code returned from cognitive service. Is there any integration I need to insert to use it?
absolutely amazing content man! Trying to build a FAQ bot based on video transcripts and after providing the answer also send the link to the source video. There's a template and video from Voiceflow that does this for public URLs - do you know any way this would be possible for private data or other data like video transcripts? I already have an airtable base with all the video transcripts and URLs that I'm looking to upload to the KB. If you have some pointers that'd be awesome but either way amazing content - will be binging over the weekend! 😅
Hey! very helpful video. I've got my file ID, tag ID, attached the document ID to my KB tags, my API in my flow is successful when I test, but I keep getting 'Null' as an answer. In my API when I Send Request as a test, I don't get errors, but for output it shows null, tokens 0, model is just " " and chunks [ ] ...I'm not sure where I'm going wrong? When I run the flow as a test, I don't get any errors, but the only answer is Null. Would greatly appreciate any suggestions.
Just so you know...your content here is amazing! I'm not finding this type of information anywhere else! Keep pressing on my man and this channel will likely explode with new subs!
Every implementation needs to be adjusted to your use case. If you feel that you need something that is missing, then adjust the implementation. You can check if the answer was generated based on the knowledge base or not. I have an example of how to do it with Voiceflow ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-BLUl951vflg.html
My dude this is absolutely brilliant! ✨✨ Quick q - If I want the bot to be able to do this at any point in the conversation, can I just set this process up once and tweak the prompt? Or will I need a set AI step after every capture card to detect whether the user has given us key info in their last message? For example, if the user gives their name to us in their first response, an email address later on in the conversation and their phone number just before leaving, will I need separate set ai steps to capture all 3 pieces of info? Or can I set it up once and ask the bot to capture key info irrelevent of when the user provides it?
From what you are describing, I feel that your chatbot is poorly designed. I'd suggest that you collect the data that you need from the customer and be done with it. If you try to extract data with LLM from every message a customer writes, you will end up using Voiceflow tokens quite fast.
@@prettyflyforanAIguy It can be a good way to do things or a bad way. It all depends on the situation. Without knowing what you are trying to do and what your current setup is, I can't say if it's a good way or not.
If it was not disabled, it would try to output an answer in the "Intent" node, and that is not what you need for this setup. First, you need to understand the intent, and then, if it's a question, you use the "Query knowledge base" agent to find an answer and show it to a customer. It all depends on what kind of chatbot design you are doing and based on that you do what you need to do.
Dude this is pretty cool! Is there a way for the bot to identify and capture certain values from a sentence? For example, if a user says "I'd like to book a test drive. My name is John." Can the bot identify that the user has given his name and capture that as a variable, whenever the user gives that information?
As with everything, it depends on the situation. Sometimes you will want to use intents as with enough examples, you can train them to be more accurate, but the downside is that with Botpress you will need to write all those examples on your own (Voiceflow, on the other hand, let's you generate them with AI so that's nice and saves time). AI transitions can allow you to build your chatbot faster, as you don't need to create intents and provide lists of examples for intent matching. The downside is that it will be less accurate as you only give it one example instead of a list, as you do for intents.
Thanks for the detailed explanation, @DoneByBots. Your insights are greatly appreciated. I understand the trade-offs between AI transitions and manually created intents. For the consultant website prospecting chatbot I'm currently building, I need to maximize accuracy. Therefore, I'll go ahead and use manually created intents, with the help of AI, of course. This approach seems to align better with my project's requirements. Thanks again for your guidance.
This is the best and easiest to understand Botpress beginners tutorial on RU-vid. I know, I have watched dozens already. The volume is a bit low unfortunately, but it's worth using headphones to make viewing easier. Good job "Done by Bots"
Exactly what I was searching for! Thank you for that Video. I have question: Currently my Bot has a Choice Language button und after that an execute code to change the language for the bot. Is there anyway to include your way aswell even tho my bot starts with the language? Changing the time out? Because when I text the bot after ending the conversation it obviously goes back to asking me what language I want.
Under Chatbot Settings there is "User variables". Documentation says that user variables never expire, so you could try setting language value there the very first time you ask your user for language selection. Each time a conversation starts, before asking for a language, check if it is already set, and if so, skip the language question. Should be easy to do.