Hi everyone! As we continue to host SnapLabs Office Hours, provide demos of SnapLabs/SnapGPT, and generally just engage with our customers and potential customers, we are taking note of questions that come up frequently. We will continue to share what we can by expanding this thread. Feel free to add any of your own questions as a reply so we can answer them too if they aren’t already here!
SnapLabs is a separate SnapLogic environment from the ones you may already be familiar with - Elastic (https://elastic.snaplogic.com) and UAT. It is intended to be a playground of sorts where SnapLogic will invite existing customers to test new snaps, features, and functionality that may or may not make it into production (aka Elastic pod). Our initial release of SnapLabs in May 2023 has focused on SnapGPT.
Note: Because it is a separate environment, SnapLogic customers that use SSO are not restricted from accessing this environment and can sign up to be on the waitlist.
At this time we use the ChatGPT 3.5 Turbo model and we are already working on the transition to ChatGPT 4.0. One significant difference between ChatGPT 3.5 and 4.0 that is a factor with SnapGPT is that ChatGPT 3.5 completed training in late 2021, so it is unaware of information or events created after that time (you may have seen us ask SnapGPT “Who won the World Series in 2022?” to make this point). What this means for SnapLogic users now is that SnapGPT/ChatGPT will be unable to help build pipelines against data sources released after late 2021 (for example, an API updated after that time).
No. While we do expand each user prompt by adding a relevant information from our documentation and any pipeline information from our own pipeline catalog that we have built over the years, we do not send any of your environment data, accounts, pipelines, metadata, etc.
SnapGPT-generated pipelines are imported, so a variable level of human interaction will be needed to select the right folder for the pipeline to import to, select the right accounts, and probably verify data objects and expressions are correct.
Configuration of SnapLogic environments is on our roadmap for SnapGPT. For an AI solution to assist with configuration beyond providing just a documentation link some amount of a customer’s environment details may need to be shared, so we want to move carefully on some functionality and focus on doing it securely.
Also on our radar and something we want to tread carefully with to prevent exposure of a customer’s environment details, metadata, accounts, etc.