Hi everyone! I am trying to use the Conversational search functionality and wanted to explore the demo available in the docs: https://conversational-search-pg-essays.typesense.org Unfortunately, it doesn't seem to be working. Is there an issue, or is it a problem on my side?

There was an issue with the edge runtime in Vercel. I've posted a PR to migrate to a node runtime. In the meantime, you can check it out locally by following the instructions in the README file

@Fanis Tharropoulos thank you, I am able to run it locally. I am also using Typesense Cloud. I suppose that in order to demo the Convesrsational Search and Natural Language Search with my documents I would need to build a dedicated UI, as it seems that the Search tab of the Typesense dashboard does not support that?

Sorry, didn't get that. The demo includes a script for indexing the data from PG essays and a nextjs app for the interface itself. The only thing that you would need on top of the Typesense instance (either self hosted or cloud), would be an OpenAI API key to create and manage the LLM

@Darya In Typesense Cloud, if you add

conversation=true

and

conversation_model_id=your_model_id

to the "Additional Search Parameters" section in the Search dashboard, you'll be able to see the conversational responses inside the cloud dashboard itself

@Jason Bosco thank you for clarifying. I am using the search with

conversation=true

and

conversation_model_id=your_model_id

parameters in the "Additional Search Parameters and I get the result documents, but don't see the user-system dialogue messages in order to proceed with follow-up questions.

Could you share a screenshot of your search parameters section in the search dashboard, and also the results pane on top?

search parameters section:

@Darya There was a bug in the UI in how it was parsing responses from v27.1 of Typesense (vs later RC builds). We've pushed out a fix for this. Could you refresh the page and try again now?

@Jason Bosco Thank you, it is working now!