With TetraBot AI your team can speed up analysis to deliver meaningful insights
Analyze transcripts with TetraBot’s AI capabilities. When prompted to analyze an interview, TetraBot does so in two ways: Tags and Interview Script Questions.
When analyzing with TetraBot, it will automatically create both types of snippets, but both don’t have to be set up in order to run. In other words, a project doesn't need to have identified a question in the materials to run the analysis, as long as you have a tag with a definition or vice versa.
Automated Tagging
TetraBot will automatically identify moments in a transcript, create snippets, and apply tags to the snippet based on the tag's definition. If there is no definition, that tag will not be used in the analysis.
For each tagged snippet created, the TetraBot will explain its reasoning for using a tag and creating a snippet in the snippet note.
Automated Q&A Identification
Identify questions by clicking the TetraBot button in your Materials toolbar. TetraBot will use those questions to find answers in a transcript during the analysis stage.
When an answer is identified, a snippet will be created. This snippet will not have a tag associated with it, but it will have a snippet note with the original question.
Timing and Process
TetraBot runs many processes when analyzing long transcripts across several questions and tags. The more tags and questions that are used as input for analysis, the longer it may take TetraBot to create results. The longer the transcript, the longer TetraBot will take as well.
Things to note
Snippets will not be created for every existing tag with a definition, the TetraBot will work to create a snippet for each tag with a definition, but if it does not find an association with enough confidence one may not be created.
TetraBot will not necessarily find an answer to all identified questions in the material documents. It will attempt to, but will only create snippets it has a certain level of confidence in the association.
Snippets created by the TetraBot will have the CREATED BY field populated by a TetraBot avatar.
When adding definitions to a tag or creating questions that are designated to be used in a TetraBot analysis, the more descriptive, concise, and clear the question or definition is the better the results will be.
Questions that lack context or have context from other questions will perform less optimally because TetraBot does not use context from other questions. In other words, if there is a question that asks: “Do you like commuting to work?” and there is another question “If yes, why?”, there will likely be a poor result for the second question. This is because TetraBot is looking for an answer to “If yes, why?” with no other context.
If the moderator repeats the answer to a question or paraphrases it, it is possible that TetraBot will consider that as an answer that is better than others and will identify that to create a snippet from. The entire transcript is input for the TetraBot, not just the participant's dialogue.
How to get the best results
Have detailed, clear, and concise definitions for tags.
Have detailed, clear, and include relevant context for each question.
If extensive questioning, try initially tagging just the most important ones.
How to filter and search TetraBot snippets
All snippets created by the TetraBot will have the TetraBot as a creator so you can use the filter on the Snippets page, go to Created By, and type in Tetra.
All TetraBot snippets for automatic tagging will have tags associated. So you can filter based on the tags you are looking for.
All TetraBot snippets from automatic Q&A identification will have the question in the snippet note and will not have a tag. Search for snippet note and type in the question to filter for an answer to a specific question.
