Alex M.

Yes, spot on! I think the continuous improvement loop is the most important. In our step that produced the semantic layer, we used a large frontier model (Fable 5). To query against it, we used a smaller but still modern one. I think it is the build/improve step that really need the max intelligence.

You can also add to your prompt to favor composing queries from existing views rather than creating new ones (especially in your prompt that builds the semantic layer). Maybe even a subagent reviewer to look for similar views and combine them (like a code refactoring pass)

Howdy! Great questions! Generalization is definitely still a hard problem. In our experiment, we used a pretty representative sample of the question set. Our goal was actually to have a training question for each "domain" of the set. When going into production, I think the highest ROI is really to increase that training set so that it covers a large fraction of the types of questions that people ask. Then if someone asks a question that didn't get a good answer, you'd want a feedback mechanism where you get notified when that happens so you can add more training questions. What model were you using for your evals? It does require a fairly new model for this to work well, but it can be a small new model.

Howdy Ezio! If you are using the DuckDB UI extension, you have another option already available to you! The DuckDB UI actually stores all of the state of the notebook (including the SQL in every cell) in a local DuckDB file. So, you could use the DuckDB UI to query that data and export it in whatever format you'd like! The excel community extension for DuckDB would work for exporting to xlsx as well. That DuckDB database is stored here: ~/.duckdb/extension_data/ui/ui.db

Howdy! If you use a DuckLake database, you can detect changes using the data change feed! You can use a lightweight poll to check the latest snapshot, then if it is new, pull the changes. There isn't a concept of triggers, but you could use a MotherDuck flight to check on a schedule

On our business plan, we do have support for PrivateLink! Would that meet your needs? Happy to answer any questions!

Hi John! Can you tell me a bit more about what you mean? Do you mean IP as in intellectual property, or IP as in "IP address"? If you share a bit more about your use case, I may be able to help!

Any time! Let us know how things go - we'd love to keep you unblocked as well as document how to use Hypermode with MotherDuck!

That last error seems to indicate an issue with your token. Is it possible that it is not in the correct format? Also, when passing it in, sometimes I need to wrap it in double quotes on the command line. Maybe using '"mytoken"' would work?

Ah, so, how can I help you eliminate that copy/paste workflow? If Hypermode can run Python for example, you can connect to MotherDuck in Python and execute those queries!