This release includes the addition of a server which allows developers to interact with their dbt project independent of the RPC server and in a much more performant way. The interface is also greatly improved. This is intended to be used as a development server -- though it could scale horizontally in theory.
The interface simply involves POST requests to either /run or /compile with a content type of text/plain containing your SQL with as much or as little dbt jinja as you would like. The result is an enum response of a dict containing top level key errors on failure, otherwise a result with top level key of result for compile endpoint and top level keys column_names, rows, compiled_sql, raw_sql for run endpoint.
This integration is fully compatible with dbt power user datacoves fork here which adds extended functionality including model preview panel that can execute a dbt model and show results in a web view with a button or even more conveniently, with control+enter like in many SQL IDEs.