This workflows demonstrates how to summarize meeting transcripts with models hosted in the KNIME Hub GenAI Gateway.
GenAI is also used to generate example meeting transcripts since actual meeting transcripts are sensitive information that should not be shared on a public place like the KNIME Community Hub.
Despite the fictional nature of the transcripts in this workflow, the general summarization approach will also carry over to actual meeting transcripts.
Workflow Details
The first part of the workflow connects to the GenAI Gateway of a KNIME Business Hub and selects a model. This requires to authenticate with the KNIME Hub Authenticator and then select a chat model registered in the GenAI Gateway. For this example we used OpenAI's GPT-4o but other models from other providers can be used as well.
In the second part, the chat model is used to generate meeting transcripts of various types. Double-click the Generate Meeting Transcript component to change the type of meeting. You can also explore the component to learn how to generate a meeting transcript with multi-step prompt engineering.
The third part of the workflow extracts a summary from the meeting transcript which includes a general summary, the attendees, action items and topics for a follow-up meeting (if applicable). Other kinds of information can be added by adding them in the system prompt of Chat Model Prompter.
The final component in the workflow formats and displays the generated summary.