AI that understands your workflow

AI is becoming increasingly integrated into developer tools, but the key to making it truly useful lies in its ability to understand context. Context-aware AI goes beyond generic assistance - it anticipates user needs, adapts to specific workflows, and reduces friction in complex tasks. This is especially valuable in Infrastructure as Code (IaC), where deployments are often intricate, error-prone, and dependent on various environmental factors.

In this case study, I explore how context-aware AI could transform the deployment experience within Pulumi, helping users navigate cloud infrastructure more efficiently and effectively.

Why Pulumi?

The Pulumi product provides a great canvas to explore new way to define and manage deployment needs for users who are frustrated with navigating complex AWS and Google Cloud web deployment interfaces.

What is a Deployment?

In web development, deployment refers to the process of releasing a web application or service from a development environment to a live (production) environment where users can access it.

With Infrastructure as Code (IaC), deployments go beyond just uploading code—they involve automating the provisioning and configuration of the underlying infrastructure. Instead of manually setting up servers, databases, and networks, IaC allows developers to define these resources in code using tools like Terraform, AWS CloudFormation, or Ansible.

My Work

I spent about 2 days of working time on this - both understanding the infra-as-code problem space and creating an AI solution that might materially help anticipate users needs and get them to a finished and custom deployment quicker.

Problem Statement

Deployments can fail for many reasons, and users often need to rule out each possible cause. WIth familiarity and experience, this is quicker, but it is rarely easy even for the most experienced users of common cloud providers.

Proposed Solution

Context-Aware Pulumi AI can help users deploy their preferred environment successfully.

User Analysis

I created three supposed personas based on my research and understanding of the infrastructure deployment space and use cases.

User Flows

I focused on the Developer Divya persona to create two flows showing what the best-case AI interaction could look like.

Flow 1 | Set up and deploy with a new cloud provider

Flow 2 | Add Resources to an existing deployment

In Product Mockup

Finally, the following mockup shows what this functionality might look like in context of the Pulumi product.

What this AI functionality might look like in the existing Pulumi UI.