How Duet AI for Developers works
One of my interest and focus areas in 2024 is to introduce Duet AI to developers and help them use it. It often starts with an Introduction to Duet AI deck along with a few demos.
In multiple presentations done this year to the community, customers and individual developers have resulted in a common set of questions. These questions can clearly be classified as Enterprise specific requirements or needs that developers foresee before the tool can be adopted into the Enterprise.
Some of the questions are:
- Does Google Cloud store the prompts that are sent to the Duet AI service?
- Does Google Cloud use the prompts for any training?
- Where is the Duet AI service located? Where are the results served from?
- What guardrails are possible (input and output) while sending/receiving data from Duet AI?
- What codebases has Duet AI model been trained on?
We cover the above questions and much more that you can access and cite from our official Duet AI documentation page. Here are specific documentation links for your reference too:
- 📙 How Duet AI in Google Cloud uses your data
- 📙 How Duet AI helps you generate code and cites sources | Google Cloud
- 📙 Duet AI in Google Cloud and Responsible AI
- 📙 Certifications for Duet AI | Google Cloud
- 📙 Duet AI locations | Google Cloud
Additionally, there is a very useful section in the documentation that tries to capture the content in the above links in a summarized form and which can be useful to share with the stakeholders in your organization. The section is titled “How Duet AI for Developers works” and I reproduce the screenshot here.
Hope this post helps to address several of the queries that you might have around Duet AI in Google Cloud, especially within an Enterprise environment.
Let me know in the comments if you’ve got additional questions.
