The management of infrastructure is transforming, with AI agents increasingly taking center stage. Pulumi, a key player in this field, recently unveiled a suite of new capabilities tailored for what it calls the 'agentic infrastructure era.' Just a year ago, discussions about AI agents and infrastructure were nearly absent, according to Pulumi's CEO Joe Duffy. Today, AI agents reportedly handle 20% of operations on Pulumi’s platform, and Duffy anticipates that could rise to 100% in the future.
New Features for AI Workflows
To integrate AI agents into workflows, Pulumi is launching several innovative features. A notable addition is the introduction of free, ephemeral Pulumi Cloud accounts that AI agents can create independently, bypassing traditional signup processes. These accounts will expire after 72 hours, but humans can later claim them to extend their usability. This initiative aims to provide agents with easy access to the tools they need.
Additionally, Pulumi is releasing an npm package that enhances the command line interface (CLI), allowing for one-shot invocations. With the command npx pulumi <anything>, agents can initiate operations without the usual installation hurdles. This improvement is complemented by a new imperative command, pulumi do, which enables agents to provision cloud resources more efficiently. For example, the command pulumi do create eks:Cluster creates an Amazon EKS cluster directly, removing the need for agents to navigate complex project structures or dependencies.
Enhancements to the CLI
The CLI is also being upgraded with over 30 features that were previously exclusive to the Pulumi Cloud web console. These enhancements include change history, drift detection, audit logs, and secrets management, all developed over the past three years. Duffy compares these upgrades to the functionality of GitHub’s gh CLI, which agents seem to prefer, indicating a shift toward more user-friendly command line tools for AI-driven tasks.
The new CLI supports JSON output and structured error messages, simplifying how agents process responses and integrate them into their workflows. These features are essential, enabling more sophisticated interactions between AI agents and cloud infrastructure.
The Future of AI in Infrastructure
As the demand for AI-driven solutions grows, the infrastructure sector is set for rapid evolution. By 2025, it's expected that AI agents will dominate operations within platforms like Pulumi and reshape infrastructure management across various industries. Duffy’s insights reflect a broader sentiment in the tech community about the need to adapt to these changes.
Looking back on this transition, Duffy noted that initial skepticism around AI agents in infrastructure has shifted to hopeful anticipation. "They were kind of like, we have no hope. And then we said, well, we’ve got this thing, Neo. Without the AI, they probably had no hope," he remarked, highlighting the significant change in mindset.
As Pulumi leads the charge into this new era, it sets a standard for other companies to follow. The implications of these advancements reach far and wide, potentially influencing everything from compliance fixes to global policy management. With hundreds of thousands of IAM policies often needing updates, the ability to implement changes seamlessly through AI could transform the industry.
The integration of AI agents into infrastructure management is more than a passing trend; it signifies a fundamental shift in how operations will be carried out in the future. As more companies recognize this potential, the infrastructure sector will undoubtedly continue to evolve, driven by the capabilities of AI.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
