Summary
We are thrilled to announce the AI Copilot (Preview) for QMCLOUD which is an assistant that can automatically create the appropriate infrastructure based on a user entering simple descriptions. The AI Copilot capability in conjunction with the already available No Code technologies (Drag, Drop, Smart Wizard, Auto Connections, Templates) will further drastically reduce the effort required to compose and deploy cloud infrastructure.
Details
The AI Copilot is integrated into QMCLOUD using the Azure Open AI (AOAI) service in the backend. The current version uses the gpt-35-turbo language model. We are planning to release a newer version with the gpt4 models in the coming months.
The QMCLOUD platform passes the user descriptions to the AOAI service and receives either a program or a JSON configuration file that is compatible with the QMCLOUD canvas. The files are then processed and are rendered on the canvas similar to an architecture diagram.
Use Cases, Advantages
One of the main use cases for the AI Copilot will be where the user is unsure about all the components required to deploy a service. For example, in AWS EKS, a user will need to compose many components including VPC, Subnets, Routing, Gateways, EKS Cluster, Nodes and IAM roles and policies. The user can enter a description in a prompt and incrementally build the required infrastructure.
As with any generative AI based technology, the more information and context that is provided to the assistant, the more accurate responses will be. The advantage of QMCLOUD AI Copilot is that the assistant will render the desired infrastructure on a canvas where the user can review, make changes if required and validate before deployments.
Example deployment
The following are the main steps to deploy any infrastructure:
1) When you create a stack in QMCLOUD, you can enter a description of the desired cloud infrastructure you want to deploy. For example, in the AI Copilot assistant prompt, you can enter the following description to deploy an AWS EKS (Elastic Kubernetes Service) as shown below:
2) The AI assistant will automatically create an appropriate configuration and prompt you to confirm as shown below:
3) The AI assistant then renders the proposed infrastructure (visualization) on the canvas. The canvas will show the proposed AWS components, their dependencies using appropriate connections and their properties as shown below:
4) The user can now make appropriate changes if any and deploy the infrastructure. The modifications can be made using the various No Code technologies in QMCLOUD including drag and drop, auto connections, direct property edit.
5) Once you are satisfied with the desired state, you can deploy using push button deployments. The following diagram shows a fully configured and deployed EKS infrastructure. Note that the user has made few changes to include policies etc.
We will soon be releasing the following ready to use descriptions / prompts so users can easily use them as templates and make modifications to either the prompts or the canvas state when required:
a) AWS and Azure landing zone components
b) AWS EKS and Azure AKS Infrastructure
c) AWS PaaS based web site deployment
In summary, the AI Copilot capabilities will further drastically reduce the effort required to deploy cloud infrastructure.
You can get started with QMCLOUD with the following options:
SaaS platform at - https://app.qmcloud.io
AWS Marketplace appliance - AWS Marketplace: QMCLOUD (Enterprise Edition) (amazon.com)
Contact us at info@qmcloud.io for other self-hosting options
You can also review our short video on AI Copilot here:
