From 4a20cad046267b47c800b5d73bd44f029ad22ea0 Mon Sep 17 00:00:00 2001 From: "Leona B. Campbell" <3880403+runleonarun@users.noreply.github.com> Date: Wed, 19 Jul 2023 13:29:59 -0700 Subject: [PATCH] Apply suggestions from code review --- website/blog/2023-07-17-GPT-and-dbt-test.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/website/blog/2023-07-17-GPT-and-dbt-test.md b/website/blog/2023-07-17-GPT-and-dbt-test.md index dc2f478e934..f55b36cd071 100644 --- a/website/blog/2023-07-17-GPT-and-dbt-test.md +++ b/website/blog/2023-07-17-GPT-and-dbt-test.md @@ -1,6 +1,6 @@ --- title: "Create dbt Documentation and Tests 10x faster with ChatGPT" -description: "You cangit s use ChatGPT to infer the context of verbosely named fields from database table schemas." +description: "You can use ChatGPT to infer the context of verbosely named fields from database table schemas." slug: create-dbt-documentation-10x-faster-with-ChatGPT authors: [pedro_brito_de_sa] @@ -18,7 +18,7 @@ By now, everyone knows the wonders of the GPT models for code generation and pai -As a one-person Analytics team at [Sage](http://www.hellosage.com/) I had to create our dbt pipelines from the ground up. This meant 30+ tables of internal facts and dimensions + external data into a Staging Layer, plus all of the following layers of augmented models and Mart tables. After the fact, we are talking about 3500+ lines of YAML that I was NOT excited to get started on. Fortunately for me, this was February 2023 and ChatGPT had just come out and boy, was I glad to use it. After a good dose of “prompt engineering” I managed to get most of my documentation and tests written out, only needing a few extra tweaks. +As a one-person Analytics team at [Sage](http://www.hellosage.com/) I had to create our dbt pipelines from the ground up. This meant 30+ tables of internal facts and dimensions + external data into a Staging Layer, plus all of the following layers of augmented models and Mart tables. After the fact, we are talking about 3500+ lines of YAML that I was NOT excited to get started on. Fortunately for me, this was February 2023 and ChatGPT had just come out. And boy, was I glad to have it. After a good dose of “prompt engineering” I managed to get most of my documentation and tests written out, only needing a few extra tweaks. Writing this article as of July 2023, and now powered by GPT-4 and not GPT 3.5, it is already easier to get the same results I did, so here are my learnings that I hope everyone can replicate.