These are some things that I noticed could be improved on the new (amazing) Studio interface. I know it’s a bit extensive, but I would love feedback on each item (feasibility or dev status).
General
There should be labels on cards and in the flow editor indicating the saving status. The “last saved” info is not enough. Although I assume that all updates are autosaved now, it would still be nice to have confirmation that they’re being saved. Sometimes alterations are not actually saved, and if I reload the page, I lose them all.
There should be a toggle in message cards to instruct the AI not to change the message content. My bot has personality rewrite enabled, but I would like to keep some messages intact.
Emulator
There should be a “Go Back” button in the Emulator to return to the previous stages of the conversation. Often, I need to test a card with different values, and right now, I have to reset the conversation and start it all over again which is unproductive.
The emulator should be faster. Sometimes it takes up to 10 seconds for the messages to be displayed in the Emulator after being shown in the logs.
Style
[ ADDED] The “New Conversation” and “Start as New User” buttons should be placed in the Emulator header (perhaps as icons), so we don’t have to first click the toggle menu to access them.
The Capture Info card should display the question text when the response is not being assigned to any variable, such as “Capture Info: Do you confirm?”. Alternatively, you could remove the “Capture Info” label altogether since we can tell the card’s purpose by the icon.
Transition lines can become messy too easily. It would be helpful if they could self-organize so we don’t have to waste time making the flow editor less spaghetti-like.
[ ADDED] There could be a toggle option for auto-aligning nodes on the grid; it helps with organization. Or simply a button for organizing everything at once (nodes and lines) when you click it.
Node names should be allowed to be bigger, have special characters and not be required to have hyphens between words.
Those are fantastic feedback right there!!
However, some of them are already submitted to the team to look into, I have submitted the rest of the ideas.
I would like to test the behavior of my bot when I provide different values in the capture information cards, without having to start the conversation all over again.
For example I have a Show-Search-Results node with a Prompt card for the user to select the desired result. After selecting, the bot transitions to a Booking-Confirmation node. I want to be able to select a result, see the confirmation text, and then simply go back to the previous node to select another result.
Dialogflow CX has this function, it’s called “Undo Last Conversation Turn”. It’s like going back in time for many turns you need (they limit it to 5 times though), with all the workflow variables being reset to that point in the convo.
Hey Bassam, I thought of another (better) example of how the “Undo/Go Back” button would be useful.
Let’s suppose I am testing a conversation flow and I make a change in a card (any kind) or node, in order to see the new behavior I would just need to go back a few turns instead of resetting the emulator and having to start it all again.
Do you guys have an estimate about when it will get faster? Because at the moment it’s even hard to test the bot, since there is no way to go back easily in the conversation, nor disable personality rewrite for specific messages in order to have faster outputs.
(There even a privacy concern here because all outputs - which may include sensitive information - are being sent to OpenAI to process, sometimes unecessarily)
But I understand it’s up to OpenAI to cater to the exponential demand of their servers!
And by the way, when is Studio expected to leave preview mode?
Do you guys have an estimate about when it will get faster? Because at the moment it’s even hard to test the bot, since there is no way to go back easily in the conversation, nor disable personality rewrite for specific messages in order to have faster outputs.
It depends on ChatGPT performance . but as a workaround, what I do is, I connect the start node directly to what I want to test and have a small execute code that sets everything I need.
(There even a privacy concern here because all outputs - which may include sensitive information - are being sent to OpenAI to process, sometimes necessarily)
You have the option, you don’t have to use OpenAI-related components, like AI Task & Personality
And by the way, when is Studio expected to leave preview mode?
So we still have the “Preview Release” label because not all of the Botpress cloud features are present. We are adding more features as we speak and are still doing a few tweaks to offer a better experience to our users. But please rest assured; we have thousands of clients who have successfully deployed their bot(s) already!
ADDED] The “New Conversation” and “Start as New User” buttons should be placed in the Emulator header (perhaps as icons), so we don’t have to first click the toggle menu to access them.
. Cheers
. but as a workaround, what I do is, I connect the start node directly to what I want to test and have a small execute code that sets everything I need.