Pulumi is a critical component of the AI agent stack because it provides the programmatic 'hands' necessary for agents to manipulate cloud resources. Unlike static configuration tools, Pulumi’s Automation API allows developers to embed infrastructure management directly into their software, making it the preferred choice for those building autonomous developer agents or self-healing systems. If an LLM needs to provision a GPU cluster or adjust security groups, Pulumi offers a typed, reliable interface to execute those changes.
The company is also a first-mover in specialized agents with Pulumi Neo, which functions as an AI platform engineer. Neo demonstrates how agents can move beyond simple code generation into actual execution and reasoning about infrastructure state. For the broader ecosystem, Pulumi serves as the bridge between high-level AI intent and the complex, multi-cloud reality of modern infrastructure.
In 2017, the infrastructure industry was largely committed to declarative configuration files—static blocks of text written in proprietary languages. Pulumi entered the market with a different thesis: infrastructure is code, and it should be treated as software. Founded by Joe Duffy and Eric Rudder, both alumni of Microsoft’s developer division, the company built a platform that allowed engineers to define cloud resources using TypeScript, Python, Go, and C#. This move brought the entire software engineering ecosystem—loops, conditionals, testing frameworks, and IDE support—to cloud provisioning.
While Terraform remains the incumbent, its reliance on HashiCorp Configuration Language (HCL) creates a barrier between application developers and platform teams. Pulumi removes this wall by allowing developers to use the same languages for their infrastructure that they use for their apps. This allows for higher-level abstractions; a team can package a standard VPC and database setup as a reusable class in a private npm or PyPI registry. This approach has gained traction with over 3,700 companies, including Snowflake and Mercedes-Benz, who find that managing thousands of resources is more reliable when using standard software development patterns.
The emergence of Large Language Models has shifted the company’s focus toward what it calls agentic infrastructure. This is where Pulumi differentiates itself from legacy tools. Because its platform represents infrastructure as a program rather than a static file, it is natively compatible with how LLMs interpret and generate logic. This led to the launch of Pulumi Neo, an AI agent for infrastructure. Neo is not a simple wrapper around a chat interface. It has access to the Pulumi state and can execute tasks end-to-end within the guardrails of the platform.
Neo is designed to understand a company’s specific context, including policy constraints and past deployments. If an engineer asks the agent to optimize costs or fix a security vulnerability, it can generate and execute the plan directly. This represents a transition from human-in-the-loop management to autonomous operations, where the AI understands the intent and the underlying engine handles the state consistency.
Beyond its own agent, Pulumi provides the foundational layer for other AI agents through its Automation API. Most AI-driven development agents struggle with the final mile of cloud deployment. By providing a programmatic interface to the provisioning engine, the company enables third-party developers to build agents that can provision their own testing environments, scale clusters based on real-time demand, or self-heal during an outage.
Headquartered in Seattle, Pulumi has raised significant capital, including a $41 million Series C in late 2023. This funding reflects a broader bet that as cloud complexity grows, manual configuration becomes impossible. While IBM’s acquisition of HashiCorp and the rise of the OpenTofu fork have intensified competition, Pulumi’s unique technical moat at the intersection of general-purpose programming and agentic automation provides a distinct alternative to purely declarative tools.
Infrastructure as Code using general-purpose programming languages.
The industry’s first AI agent built for infrastructure.
Pulumi is hiring.