Our inspiration for tone In was to help users better navigate the often complex and diverse environment of Slack channels. We recognized that users can find it challenging to gauge the tone of a conversation, particularly in busy channels or workspaces with new members constantly joining. We also wanted to provide administrators with a tool to indicate the level of formality of their channels to users.

What it does

We created tone In, a Slack application that uses machine learning and data collection to educate users on the tone of their work and personal environments. By analyzing the tone of each message in a channel using OpenAI's "davinci" model, tone In provides users with real-time feedback on their message's formality and professionalism. It also measures each message against an overall channel tone or a tone value assigned by an administrator to ensure consistency.

How we built it

tone In was built using Python and integrates with the Slack API. We used OpenAI's "davinci" model to fine-tune our machine learning algorithm and develop our AI-generated responses feature. We also created a gamification feature that ranks the most and least professional users and a visualization feature that sends users a bar graph of top professional users.

Challenges we ran into

One of our biggest challenges was finding suitable documentation for the Slack API that could meet our specific needs. Additionally, having multiple AI models required extensive testing using real conversations from Slack channels, which was time-consuming due to limited token returns per minute.

Accomplishments that we're proud of

Despite these challenges, our successful implementation of tone In provides users with valuable insights into their tone and helps them communicate more effectively in Slack channels.

What we learned

One of the main challenges we faced was finding good documentation for the Slack API that was tailored to our needs. We also had to test each AI model with actual conversations from the Slack channel, which was time-consuming due to the limited number of tokens that could be returned by the AI per minute.

What's next for tone In

Moving forward, we plan to continue improving tone In by adding more advanced AI models and improving the user interface. We also hope to integrate with other messaging platforms and expand our user base.

Built With

Share this project: