Introducing Rafay's Generative AI based Copilot¶
A few days back, as part of our early March 2024 release, we opened up Rafay's Generative AI based Copilot to our customers. For the folks that are active readers of our product blogs, you will recognize that this is the result of a GenAI focused Hackathon we ran in late 2023. You can read more about our learnings from the Hackathon in 2023.
Just like Batman works way better with Robin as his copilot, we are seeing our customers benefiting immensely by using the Rafay Copilot that is integrated right in the console. In this blog, we will use a few examples to showcase the value of the Rafay copilot.
Background¶
We invest heavily on an ongoing basis to ensure our customers have a streamlined and intuitive user experience in the Rafay Platform. But, there are still questions for which administrators have to spend time reading our Product Documentation. The first milestone we targeted with the Copilot was to provide the means for users (i.e. platform teams, Ops/SRE, developers and data scientists) of our platform to get responses to questions right inside the console. Some of the criteria we wanted to address are listed below:
Zero Context Switching¶
To ask questions, users should not be required to switch context and navigate to a separate documentation site etc.
Natural Language Interface¶
The copilot should provide users with a ChatGPT style "Natural Language" interface where they can ask their questions in simple English.
Relevance¶
We update our product documentation multiple times a day. The copilot is configured to train continuously on our documentation. The copilot is able to keep up with incremental changes to our documentation and is aware of pretty much every word in our documentation.
Source Data¶
We made the decision to make sure that we display the URLs to the data sources that were used by the copilot to generate the response. This not only provides the user confidence about the response, but also provides the user with a clickable link where they can learn more.
History¶
The copilot maintains a query history for every user. This allows the user to review the question they asked before and run the query again. The copilot also provides the means for the user to delete each item in the history if they want to.
User Feedback¶
Although we have made a concerted effort to make sure the copilot does not hallucinate, there is always the chance that the responses were not ideal or helpful. The copilot provides users with the facility to provide feedback. We actively review this feedback and optimize our GenAI backend.
Real Life Example¶
When platform engineers start with the Rafay platform, one of the first things they look for is whether they can use a CLI to automate things. Let's see how this user would have used the copilot to not just answer this question, but also complete the folliwing in approximately 30 seconds.
- Download the CLI,
- Configure the CLI and
- Test it
Watch a brief video showcasing the Rafay Copilot based experience for this user.
Try this Out¶
We are rolling out the Rafay Copilot gradually to our customer Orgs. Please contact us if you would like to fast track access and have this enabled in your Rafay Org.
What Next?¶
In the next few blogs, I plan to dive deeper into the design and architecture of our copilot. I will also write a blog describing how our customers use the Rafay Platform to efficiently run "world class" GenAI hackathons internally to help bubble up compelling se cases that could completely transform their companies.
Thanks to readers of our blog who spend time reading our product blogs. This blog was authored because of the significant number of questions our customers have been asking us about how we integrated Generative AI into the Rafay Platform. Please Contact the Rafay Product Team if you would like us to write about specific topics.