Hi Cédric, I have already encountered the same kind of issue with one of our clients. In our case, it was due to Google Chrome's real-time translation feature attempting to translate the LLM outputs written in the Dust interface. To fix it, it is recommended to ensure that Google Chrome does not try to translate the LLM outputs, for instance, by asking Chrome to never translate this specific site (ie the Dust interface).
Could you give 5 examples of questions you would like to ask to meeting transcripts from D/M/Y to D/M/Y? => Our main need is to ask questions regarding a specific (single) transcription for a given client and a given date (or to a group of transcriptions, but for always a specific client). In our specific case, we therefore do not need to conduct analyses on a large number of transcriptions (a targeted rather than a global approach). Examples of question could be: What is the Document Management System used by the client ? What external data sources does the client wish to integrate into Dust? What is the client's tech stack (Microsoft, Google, or other)?
Would you like to analyze transcripts from the past 30 days, 90 days, or year? What would be the ideal maximum timeframe? => not a use-case for us.
Could you help us assess the size of the ideal data set of meeting transcripts you would like to analyze with Dust? => not a use-case for us.
What are the most valuable use cases, according to you, from the most useful to the least useful:
Other (please add yours): searching specific insight into a single transcription (for a given customer and a given date) => Ideally, the Dust Assistant should be able to select the right transcription dynamically.
Summarizing a meeting in a given format for a sales rep.
Use the data extracted from the meeting transcripts to update internal knowledge.
The 🔒[private feed] channel is really very interesting and inspiring. Suggestion => insights on the progress of the Dust roadmap would also be very useful. I am referring to features that are not yet available but currently under development, such as the connector for MS SharePoint, or the establishment of the Dust Data Center in Europe, etc. (It's also conceivable that the community could vote to express their preferences for prioritizing features.)
