AI chatbots are everywhere now, but not all of them make you want to actually use them 
What’s one thing that immediately puts you off?
That’s interesting! Those examples really do look pretty overloaded. How did you handle this with your own chatbot - with pre-made questions, or in some other way?
We follow the good ol’ acronym: KISS (Keep It Simple, Stupid). Thanks to Team Elfsight, we can customize our AI chatbot and always keep it simple-looking while offering the answers our clients or prospects expect.
Quick heads-up: we’re trying out nested replies in this discussion
You can now reply directly to individual comments - curious to see how it feels!
If you are required to choose one or several topics before being able to even ask your question - or worse, you are just sent on to the guides section without being able to ask the chat anything.
On top of that, if it’s not easy to get in contact with a real person. Chats are great (if you disclose what their actual function is), but sometimes humans are just better.
A great example of the is Squarespace’s chatbot. You have a narrow selection of topics to choose from that might not fit your problem, and then you have to insist on talking to a human, that then sends you on to a support request form. This is probably not a problem for most companies using Elfsight’s chatbot, but help should be easy to get. 
Great example, @Support_Branchekatal! I guess there’s a big difference between using pre-made questions to help people navigate and locking them into a limited set of topics with no option to just ask their question.
Is your chatbot set up to hand conversations over to a real person when needed, or do you rely solely on the chatbot?
We have a lot of information for the AI chatbot – mostly in JSON files, including images and links. However, it sometimes asks the user five or more questions, occasionally returning results, which is understandable. If the user continues asking questions, that’s fine. But the AI chatbot keeps trying to narrow down the results – that’s almost too much.
Interesting point! Sounds like the chatbot sometimes tries too hard to be helpful 
As I mentioned in another thread, I would have to say when they replace support for legal matters or more urgent tasks. Whether there is account hacking, suspicious payments, or if there is an automated error with a subscription or service. In my opinion, AI chatbots are good for more general tasks, but not urgent.
For example, I was kicked off of an AI service for an incorrect violation. I made an AI video of a girl in a bathing suit riding a giant lobster. It was not a sexual bathing suit, just a top and briefs, but the system detected a violation that ended up hurting my reputation to that service. And because of AI chatbots, there was no one I could I report it to. I explained it to a chatbot, and it was sarcastic to me.
And it’s off topic, but supposedly there will be robot police in the future, whereas I don’t like to put too much faith in automations, since errors happen. So this is what annoys me, to answer your question. I use AI a lot, but I don’t support that it should replace all human labor.
The times when you have added content for the Chatbot only to be told that the office is closed.
Another example is when you are asked to add details of the issue, press enter and then you are asked what the issue is…
Our chatbot is mostly used and intended as a ‘helpful librarian’ to look through our many PIM-system user guides, but yes it’s easy to get in contact with our support via a button at the top.
I do understand, that very large companies like Squarespace needs to try and have as few support request as possible, if they can be solved via the chat and their guides, but it can (in my opinion) also get too restrictive.
Ohh, sounds frustrating indeed! That’s a great example of a chatbot flow that creates more work instead of saving it 
Absolutely, I see your point, @Prevalent_Arts. AI can be great for handling routine questions, but there are definitely situations where human help needs to be available.
Thanks a lot for sharing your experience! Another good example of what can happen when too much is left to AI 
Absolutely! The idea is not about adding an AI chatbot and calling it a day - it’s about setting it up in a way that actually helps people. And if users start feeling restricted rather than helped, that’s probably a sign the flow needs rethinking.
“Helpful librarian” sounds amazing. I hope it lives up to both your and your customers’ expectations
Not just Chatbots per se, but I have had the following experience more than once. I need to get something. I type it into Google search to find who is selling it. Google gets me right to the product page. And then … the pop ups start. I can see what I want to buy under all the pop ups. I want to buy it. But I can’t get to it. There’s the cookie consent. OK, I get it. Then there is the request to review them. I haven’t bought anything yet. “Give us your e-mail so we can tell you about all our special deals.” No. Then the Chatbot. And then the Chatbot pops up AGAIN asking “are you sure you don’t want to chat?” NOOOOOO. I want to buy the damn thing. I will at this point just go onto the next website on the Google search that will let me actually buy their product.
Horrilarious - welcome to the community, @user31905!
That’s right - a website needs to keep visitors’ focus on a product rather than throwing distractions in their way that can actually backfire.
Funnily enough, your comment would fit perfectly into our previous discussion about website pet peeves: What makes you close a website right away?