Have something to say?
Widget should stay open when navigating
When I am using the widget as a documentation tool. When doing a navigation, the widget is reloading, you are losing the current discussion, and the bot closes. The user must reopen the widget, click messages, then select the discussion to find his following.
Display change
I would like to change the red parts to Japanese if possible. If it's not currently possible, I would appreciate it if you could make it customizable.
Duplicate Chatbots
It would help us immensely to have an option to duplicate an existing chatbot—as we need additional bots are all trained on the same core knowledge, yet each respectively having their own unique branding, persona and areas of expertise.
ability to duplicate existing chatbot when creating new one
i want to create another chatbot but need it to be trained already on everything the first one is. so just to “duplicate” and then change different characterizations.
search pages option on trained URLs list
we have hundreds of pages and they get updated sometimes and i need to know what already trained and what i need to retrain, so please add “search” in URLs list
Manage/group links & gitbook dat sources
Very long to remove multiple pages in links or from gitbook → no selection/checkbox feature or grouping of links = hard to manage In gitbook there’s no way to see the link of the page that have been loaded or easily manage order and add/remove pages (Some page have the same name so I need the link to see from which subpages it came from) In gitbook the arborescence is very important to manage knowledge and know what to refresh
typing ...
Hi there, This is a great tool. Would be great if we could have a little “…” to indicate when a user is typing
Tooltip experience
If tooltip is enabled on the chat bubble and the user decides to close the tooltip, it should stay closed in the same session. Now if a user closes it and visit a next page on the site in the same session, the tooltip is again visible. From a user experiene it would be best to keep it closed and thus not show it anymore.
Translate chat logs to review
When reviewing chat logs, it would be helpful to have a "translate" button that lets you view the chat in your default language, similar to how Instagram does it. We work globally and our chatbot speaks to people in many languages, but I can only review chat logs for English conversations.
Improved visualization options of LLM output
Now the output is purely textual with some limited styling applied. If i.e. the bot is referring to a set of products, it would be nice if this could be shown by making use of a carousel / gallery of cards like https://learn.voiceflow.com/hc/en-us/articles/9176043863309-Carousel
The ability to create conditional prompts
As an add-on to the 'master prompt' for specific context, it should be possible to work with conditional prompts. Use case: product recommendations based on a set of preliminary questions where you can instruct the LLM to follow a specific logic. Q1: if answer yes > do this, else, do that .... I've tried to set this up in the main prompt but it's not 100% reliable and accurate.
The ability to grab leads via the API
Whenever I new lead is coming in , it should be able to grab them via the API to send them over in a specific format to the client.
Ability to ignore some URLs using meta tags or HTTP header
If you provided a way to tell the website scanner to ignore some pages using a HTML meta tag or an HTTP header, it would make it easier to ignore pages when there are 1000s of them. My website has over 10000 blog posts, and most of these posts are not trainable. Provided that I had a way to tell the scanner to ignore the links using code or servers, it would save a lot of time. It could be the same way as the meta robots and X-Robots-Tag from Google, but with your own naming.
Disclaimer Option
Any option to add a disclaimer prior to using the chatbot? We're currently using the welcome message to issue a disclaimer that this is an AI-powered chatbot and that accuracy can vary, so please verify any information with a qualified expert. Maybe an option to add a disclaimer just before launching the chat could be added, instead of having it part of the actual chat conversations.
Sitemap import through XML file
Sitemap import works great through a link, would be great if we can import the same directly by uploading XML file. Very niche feature and not at all a must have, but nice to have and could be quickly implemented.
Link sorting after sitemap import
Sitemap import is a great tool, but the ensuing filtering of which links to import to the knowledge database could be enhanced by allowing us to sort the fetched links alphabetically. For example if you select all website links but you want to exclude a section, say the news articles, if we can sort the fetched links alphabetically, it's easier if they all follow each other (e.g. website.com/news/....) and you can skip them when importing the rest to the knowledge database.
Semantic instruction for Q&A Response
Right now, we can edit Q&A's to have the bot respond to a specific question with a specific response. Example: "what are your working hours" --> "our working hours are 9 to 5" , and the bot will give that exact answer every time. It would be great if we could guide the bot on HOW to respond to specific questions instead of WHAT to respond. For example, for question "what are you working hours" the answer we set would be the instruction: "check in your knowledge database and reply with our working hours." The reason for this would be to address multiple imperfect replies to multiple questions that are closely related.. instead of correcting the bot on every single question that is closely related. In other words, it serves the same function as Q&A current feature, but groups Qs and As in groups instead of a specific Q for each specific A.
Ability to attach images to chat requests
For support requests, we need a way to capture screenshots from the user
Follow up Prompt - toggle on/off
It would be great if you could enable/disable a follow up prompt. We're using the escalation follow up prompt and it seems the only way to turn it on and off is to delete it and recreate it.