This past week we had the opportunity to demo Knowtworthy at True Blue Expo in the Mars Centre in Toronto. We met a lot of great people at the event and seeing all of the interest surrounding our idea and platform was really inspiring! So thank you again to everyone who stopped by!

Upon pitching our idea to hundreds of people throughout the day, some of the most common feedback that we got was two-fold:

  1. The sentiment analysis sounds amazing, but what kind of impact will it have on my team?
  2. From audio recording to speaker diarization, how do I make sure that my meetings stay secure?

Throughout the rest of this blog post we will dive into the thought process behind sentiment analysis in a team setting and will explain how information security works at Knowtworthy.

Sentiment Analysis and revolutionizing meetings

Very early on during our research and while conducting interviews with business professionals, we found that a significant constraint affecting teams during meetings was the team dynamic itself. Some people had conflicting personalities and some people would naturally contribute more than others and potentially interrupt, while still others would remain silent the entire duration of the meeting.

In our view - everyone at the meeting should be there for a reason and everyone's opinion matters, but often it is hard to tell someone directly what they are doing wrong during a meeting and that they shouldn't be behaving a certain way. This intervention can create additional tension which then repeats the cycle of suboptimal team performance. Our goal then, was to find a way to objectively analyze a team during a meeting in a scalable way – and more importantly – provide actionable feedback.

For this reason we decided to move forward with a sentiment engine to process the speech of individuals throughout the meeting and return a list of performance metrics. The key here is that both sentiment and attributed voice transcription are individually opt-in features and that all sentiment reports would be sent directly (through email) to the person to whom they pertain. No middle man, no manager overseeing, no teammates looking into your results, just helpful feedback for you to personally reflect upon.

With this feedback in hand, it is up to individual team members to take action upon it. For example, if it is identified that your aggression levels are far above average, you can take this into consideration and actively try to listen more during the following meeting. Alternatively, if you opt in to the sentiment evaluation but find that your reports are empty, this indicates that you spoke less than 300 words that meeting, and we cannot, in fact, provide tips since the data set is too low. This can be interpreted as a call-to-action to try and participate more in the future.

But why not share sentiment data with the whole team, or at least the meeting chair or manager? Wouldn't it make sense to provide this information directly to the person most capable of overseeing change? The reason for why sentiment data is kept entirely confidential stems from our concern about meeting privacy overall. This encompasses everything from how we store meeting data to what is done to audio and transcription data after the fact. It is both intuitive and supported by our research that individuals who know what information is collected about them and have power over it's creation and deletion feel significantly more comfortable during meetings than those who are not similarly informed. Hence, sharing sentiment data could potentially have an effect similarly detrimental to publicly scolding someone for their behaviour during a meeting. It is for this reason that we elected to keep all data as 'need-to-know' as possible.

The security of your meeting data

It is not news to anyone that company meetings need to be kept within the walls of the organization. Sharing excess information is frowned upon both implicitly in corporate culture and explicitly in corporate contracts, so every precaution must be taken to ensure the security of meeting data. A serious topic of this magnitude requires equally serious action from the Knowtworthy platform. As a result, the following features are implemented:

First, all data handling on our platform is tested and encrypted. We are committed to protecting your privacy and the privacy of your team at Knowtworthy.

Second, all elevated processing features (ie. speech-to-text transcription, speaker diarization, and sentiment analysis) are kept as 'opt-in'. So, if a team feels uncomfortable with any data leaving the room for processing, no data will. Furthermore, if speaker transcription is enabled, speaker diarization will only identify users that upload a voice print through the Knowtworthy platform. So if team members are not interested in receiving sentiment reports and would not like their name attributed to the transcripts, then simply never uploading a voice print or disabling access to it through the settings page will prevent speaker recognition for that specific individual. Detailed and diarized transcripts can be incredibly useful for keeping meeting information handy and helping to write good minutes, but if this compromises the openness of people partaking in the meeting then it doesn't make sense to use the feature.

Finally, all information stored on the Knowtworthy platform, can be deleted. Permanently. This refers to all meeting minutes, all meeting audio data, all meeting transcripts, and all meeting sentiment profiles. On top of having to opt in to several features that enable data storage, the option to delete is always available and easily accessible.

Next Beta!

The last announcement is that we are working hard on finalizing the sentiment analysis engine and are planning to release the rest of the platform to beta within a month!

You can already use the platform to keep all of your meeting info in one place and instantaneously share meeting minutes, agendas, and action items and you can sign up for that here. The new beta will be backwards compatible and will integrate will all current accounts so all profiles will be conserved once we update the platform.

If you are as excited about real-time transcription, speaker diarization, and sentiment analysis as we are, sign up to our beta list to get notified when we release!

Thanks for stopping by! If you have any questions you can always tweet at us @knowtworthy.