Generative AI & Visual Development for Cloud Journey

In our previous blog, we highlighted the need to develop a tool kit (Persistent Cloud Automation Stack – PiCAS) made of reusable, composable Infrastructure as Code (IaC) artifacts for accelerating clients’ cloud journeys while providing high-quality, compliant, governable, and observable cloud infrastructure. Although this tool kit increases time to market by 40% and lowers resourcing burden, the development teams still need to possess a good understanding of IaC.

In this blog, we will discuss how a quantum leap can be brought about in developing and deploying high-quality cloud infrastructure by leveraging technologies such as Generative AI (GenAI) and Low Code No Code (LCNC) visual development.

Challenges for Developing Cloud Infra

Over the past few years, cloud infra development has shifted to IaC. Programmatic scripting languages such as HCL, CloudFormation, ARM, etc. are used to ‘write out’ the required cloud infra services, configurations, IAM/security policies, etc. However, ‘coding’ remains a predominantly manual process that requires developers to configure an IDE with plugins, possess knowledge of multiple scripting languages and have a working understanding of the cloud. Even with access to a toolkit like PiCAS, the developers still require a few years of IaC experience.

Generative AI Tools for Cloud Infra Development

With the advent of Generative AI, tools such as Codepilot, AWS CodeWhisperer, etc. are rapidly penetrating the market. These tools allow developers to prompt or instruct the AI on the desired functionalities to be added to their programs. The AI is trained on large volumes of code to produce the functional blocks in the chosen programming language, which are then further refined by the developers for optimal performance. These tools are said to increase developer productivity by 50-60%^1,2. The developer still has to have enough programming and syntactical knowledge to correct the code that the AI has generated. Also, these tools are not trained on specialized artifact repositories/composable toolkits such as PiCAS that offer best practices right out of the box. Therefore, since these tools cannot provide prompts that use PiCAS or equivalent artifacts, special efforts have to be taken to train these tools on artifact repositories.

Persistent’s PiCloud Platform for Managing DevOps

Cloud infrastructure teams must also focus on managing the DevOps cycle. Setting up DevSecOps pipelines to scan developed code, pushing code to repositories and multiple environments, etc. takes substantial manual effort. At present, the market lacks tools that can seamlessly integrate user-friendly developer environment with required DevOps functionalities. This is where Persistent’s PiCloud Platform comes in as a one-stop alternative for ultra-easy, high-quality cloud infrastructure development plus DevOps.

Cloud Migration Market and Requirement of Skilled Resources

On the business front, the cloud migration market stood at $114 Billion at the end of the year 2022 and is slated to grow at 15% CAGR to hit $220 Billion by the year 2025^{3, 4}. The share of cloud infrastructure transformation work will be $29 Billion and migration management will be $44 Billion by 2025^{3, 4}. Globally, many companies will need their cloud development teams to rapidly build and deploy high-quality cloud infrastructure for them and their clients. However, building effective cloud infra teams in today’s world requires hiring highly skilled resources who are good at cloud technologies and IaC. Unfortunately, there is a dearth of such skills, and this significantly affects project schedules. So, how can we reduce this skill dependency further and help the industry ramp up cloud development efforts by more than 2x? The answer lies in integrating radical new technologies like Generative AI and visual Low Code No Code interface atop IaC.

PiCloud Platform Leverages Generative AI + Visual Low Code No Code Interface to Accelerate Cloud Infra Development

Our PiCloud Platform is the next-generation evolution of PiCAS. PiCloud allows developers to describe what infrastructure they want to build using simple English, as shown in the following steps:

The GenAI asks the user a few questions and fine-tunes the cloud infrastructure that needs to be architected.
It then pre-renders this generated cloud infrastructure on a visual LCNC development pane. A prompt AI layer present in the tool hand holds the user on how to edit and update this pre-rendered cloud architecture. Based on the prompt, the user uses widgets on the visual development pane to update the infrastructure. This pane has a dropdown menu to search and select cloud-specific PiCAS terraform artifacts.
The user selects the appropriate service – artifact block and drags and drops it on a visual development pane. The user can drag and drop multiple blocks for services they would like to add to the pre-rendered architecture diagram. Blocks can be interconnected by dragging connection lines from one block to another. The connection lines can be edited to map input/output variables between interconnected blocks.
The user can also click on service blocks to edit configuration parameters using a form that opens up on the screen. Using this visual LCNC pane, the user then finetunes the cloud architecture.
In the back end, PiCloud adds/updates code for composable PiCAS terraform artifacts and creates together a reusable, compliant, observable, governable and deployable terraform script.
The user can then download this reusable terraform script and deploy hardened cloud infrastructure to their target environment.

The user does not need to type in or have in-depth knowledge of terraform IaC to use this tool or use this tool to build their own cloud artifacts. In the Beta version of the tool, the user will be able to download generated terraform script and deploy it manually.

At Persistent, we utilize latest technologies like Generative AI and LCNC platforms to help you lower project costs and enhance efficiency through rapid cloud infrastructure migration/modernization. The upcoming version of PiCloud will have DevOps functionalities such as DevSecOps pipeline for code check, push to repository, push to environment, cost calculator, infra import, etc. Users can use this tool through a SaaS model. In the long run, we will evolve this tool to have governance capabilities to close the loop and provide a true one-stop-shop experience to clients looking for cloud infrastructure solution.

For more information on PiCloud or PiCAS and to understand how these platforms can help accelerate your Cloud Migration journey, please reach out to us.

References

Author

Dr. Sujay Phadke

Principal Engineering Manager

sujay_phadke@persistent.com

Leveraging Generative AI + Visual Development to Accelerate Cloud Journey