Feature requests

We would like statistics to be separated from the GDPR retention period as it applies within GDPR compliance. It is neither necessary nor desirable for us to view individual conversations. However, we do want to gain insight into the number of conversations per employee and the number of conversations that were made with the chatbot.

We would like to be able to split the opening hours per weekday. from 09:00 - 12:00 and 13:00 - 16:00, for example. For now, this is not yet an option. So we don't have to manually interrupt the live chat functionality every day.

We’d like a central screen displaying real-time chat statistics to keep our customer service team informed and agile. The goal is to gain immediate insight into performance and respond quickly to ongoing situations. Below is a summary of the key data points and features we want to see: Agent Activity: Which agents are currently logged in Number of conversations currently in each agent’s inbox Number of conversations handled by each agent today Conversation Statistics: Total number of conversations handled today Number of currently open conversations Number of unanswered conversations Number of conversations handled by the AI Agent Number of conversations escalated to human agents Channel Insights: Breakdown of conversations per channel

We would like to be able to split the opening hours per weekday. from 09:00 - 12:00 and 13:00 - 16:00, for example. For now, this is not yet an option. So we don't have to manually interrupt the live chat functionality every day.

Prevent music from stopping playing when you send a message, for example: If you're listening to music via an iPhone (speaker or bluetooth), the music pauses when you're in safari and sending a message. Inconvenient for the UX.

When entering a few URLs, you run the risk of entering some of them twice. I would like to be notified if I do this accidentally.

Some expectation management in the channels WhatsApp, Facebook and Instagram that the AI agent is actually thinking or writing. Similar to what you have in the Chat Widget

Ask the user in the end of the conversation, if the conversation was useful. They can then upvote an downvote. If they downvote there should be the option to provide more detailed feedback. There should then be an analysis in the dashboard how many people upvoted and downvoted and there should be an option to also read the feedback.

The option within these channels to adjust the Generated by AI message, similar to how you can do this for the Chat Widget

Watermelon apllies vendor lock-in, as it's based OpenAI. We want to use other models as well for the agents. This is to avoid bias, data-conformity, EU-act conformity and so on. Also agents need a fallback model in case Sam Altman blows up OpenAI ;)