Converse AI is a product that helps the HR / Management team in enforcing corporate values among their employees.

Inspiration

After COVID, There is a massive drive in the adoption of virtual meetings. With an increase in virtual meetings, there comes a concern about their quality. A slight misunderstanding can lead to disastrous effects on a business. A combination of Hate speech, Emotional and Behavioural detection and analytics in a meeting will significantly help the management to understand client satisfaction and employee behaviour.

As a famous proverb says, "Prevention is better than cure" Converse AI actively monitors meetings across the company and provides solutions to various problems the HR team faces.

  1. Converse AI instantly notifies the HR/Management team of any hate comments detected in a meeting involving clients. Early detection gives an upper hand in mitigating any issue before it becomes a severe escalation from the client side.
  2. It helps to understand employee mindset and client satisfaction.
  3. Sexual Harassment can be proactively controlled.
  4. Positive behaviour shown by an employee with colleagues and clients can be rewarded.

What it does

We are sending the chime virtual meeting transcription to Converse AI and with the help of the Expert.AI APIs, we can able to determine the emotions and the hate speeches on the meeting transcript.

The Converse AI uses Open search and Kibana virtualization to easily search and analyse the meeting quality and performance of an employee in a meeting.

In case of any hate speech, the Converse AI uses the slack message to send real-time updates to the HR and Leadership teams.

How we built it

The application is built on top of the AWS Chime Rightstart Kit deployed in AWS. Converse AI has the following features :

  1. Seamlessly integrated to the chime virtual meeting tool, it automatically invokes Expert AI endpoints to detect Hate Speech, Emotional and behavioural traits.
  2. Realtime slack integration for immediate notification when Hate Speech detected
  3. Realtime dashboard for detailed analytics using Open search and Kibana
  4. Cost-effective and Highly Scalable solution leveraging serverless AWS resources

Challenges we ran into

  1. Updating Chime source code to call Expert AI endpoints. - This is solved by monitoring chime events to call Expert AI endpoints
  2. Slack developer endpoints have a limited throughput (messages that can be sent per second) - This is solved by implementing exponential retry

Accomplishments that we're proud of

We are proud to have created a business solution that will improve the virtual meeting qualities while using most of the possible Expert.AI APIs for this use case.

What we learned

We have learned how NLP impacts a business, especially with out-of-the-box tools provided by Expert.AI reduces a lot of development effort. Also, we experienced the integration of multiple services like Chime, Slack and Kibana dashboards

What's next for Converse AI

  1. Expand functionality to other meeting services like Webex, Zoom etc
  2. Leverage Expert.AI's document processing APIs to process meeting attachments

Built With

Share this project:

Updates