AI Business · · 6 min read
Can Pulumi's AI Agents Really Automate Software Delivery and Fix Cloud Infrastructure?
We explain how Pulumi's new AI agents aim to automate software delivery and infrastructure management, and what this means for the future of engineering teams.
On Wednesday, November 18, 2026, a cloud infrastructure startup called Pulumi declared its current fiscal year to be "the year of absurd ambition." This wasn't just a catchy slogan; it was a prelude to a significant launch.
The Seattle-based company unveiled two major products, Pulumi Delivery and Pulumi Agents, designed to let artificial intelligence (AI) agents take over much of the work in getting new software running and fixing cloud systems, GeekWire reported.
What makes this particularly striking, according to GeekWire, is that Pulumi achieved this while actually shrinking its workforce. The company now has about 100 employees, down from approximately 130 last year. Yet, its annual recurring revenue per employee has more than doubled.
Pulumi CEO Joe Duffy told GeekWire that AI has reshaped roles and boosted efficiency, allowing the company to build products it had aspired to for years. It suggests a future where fewer people, empowered by advanced AI tools, can achieve far more.
But the real story isn't just about Pulumi's internal efficiency or its bold claims. It's about what this "absurd ambition" means for every business that relies on software and cloud infrastructure.
If AI agents can truly ship code and fix complex cloud environments, what does that imply for your engineering teams, your operational costs, and even the very definition of a software development lifecycle? That raises three questions.
Question 1
What Exactly Did Pulumi Launch, and Why is It a Big Deal for Software Delivery?
Pulumi's new offerings, Pulumi Delivery and Pulumi Agents, aim to automate the entire software release and infrastructure management process. Pulumi Delivery enables engineers and AI agents to code the steps for software releases, covering everything from testing and human sign-off to the final rollout, as GeekWire explained.
A key feature within Pulumi Delivery is an AI reviewer that assesses the risk of proposed changes. Using models like Jev from TypeSafe AI, it can estimate the odds of specific risks. If a risk is higher than a team's predefined threshold, the release stops, requiring human intervention. Daniel Gafni of TypeSafe AI noted that even the decision to ship code can now be made by a model. But how does a machine decide if a change is 'safe'?
Pulumi Agents, on the other hand, are designed for ongoing operational tasks. These agents can investigate outages, identify ways to reduce cloud costs, and resolve security issues automatically. Each agent operates with defined limits on its actions, and all proposed changes appear as code for team review, ensuring transparency and human oversight, according to GeekWire and Google Search research notes.
For context, Pulumi was founded in 2017 by Microsoft veterans Joe Duffy and Eric Rudder. Their core idea was to let engineers manage cloud infrastructure by writing code in everyday languages like Python and TypeScript. Duffy admitted to GeekWire that this foundational bet turned out to be "lucky timing" because AI agents excel at generating and understanding precisely that kind of code.
The significance here is that many AI coding tools stop once code is written and reviewed. Pulumi's new products push beyond that, covering the entire post-merge process of testing, releasing, and even fixing software and infrastructure. Duffy called this an "industry-changing moment" to GeekWire, positioning Pulumi to solve a critical problem.
Question 2
How Do These AI Agents Actually Work to Manage and Fix Cloud Infrastructure?
The core of Pulumi's system involves AI agents like Claude Code, Codex, and Pulumi's own Neo. These agents are trained to write and understand code in common programming languages, allowing them to make real infrastructure changes autonomously, GeekWire reported. How does this compare to traditional infrastructure management?
Pulumi’s Neo agent, for example, which debuted in September 2025, can now use models from OpenAI, Google, or Anthropic. It acts as an "extension of your team," capable of handling infrastructure provisioning, security, and compliance tasks. Beta users of Neo have reported delivering 10 times more infrastructure, deploying 75% faster, and reducing policy violations by 90%, according to Google Search research notes.
For an everyday analogy, imagine you run a busy toll plaza. You have a smart, efficient system that automatically collects tolls, identifies vehicles, and flags any anomaly. You set the rules: "Collect Rs 50 for cars," "Flag trucks over 10 tonnes for manual inspection," and "Alert me if daily revenue drops by 10%." The system then autonomously manages traffic, collects revenue, and flags bigger, riskier decisions for your final approval. Pulumi's AI agents work much the same way, operating within parameters set by human teams, and escalating when necessary.
These agents can be triggered by alerts from monitoring tools like PagerDuty, Datadog, or Honeycomb, or they can operate on a schedule or on demand. They then report their actions and findings back to teams via communication platforms like Slack, GeekWire noted.
Crucially, Pulumi allows customers to "bring your own bot." This means you can integrate your preferred AI agent, such as Claude Code, and run it within Pulumi's cloud. This flexibility ensures businesses aren't locked into a single AI model but can leverage the best tools for their specific needs. And that brings us to the third question.
Question 3
What Does Pulumi's Strategy Mean for Your Business and the Role of Human Engineers?
Pulumi's move signals a profound shift in the software development lifecycle (SDLC). Traditionally, engineers would write code, review it, and then use various tools like Argo CD or GitHub Actions, alongside custom scripts, to deploy and manage it. Now, much of that post-code writing work can be automated by AI agents, as GeekWire's reporting suggests.
For your business, this could mean significantly faster deployment cycles, reduced operational costs, and improved reliability. If AI agents can automatically detect and fix outages, or proactively suggest cost-saving infrastructure changes, it frees up human engineers to focus on higher-value tasks, like innovation and strategic planning. So, what does this mean for the future of your engineering team?
Joe Duffy highlighted to GeekWire that Pulumi is hiring "product-minded engineers who can think from our customers’ point of view, who can think like engineering managers, commanding a team of agents." This suggests a future where the engineer's role evolves from being primarily a coder to more of an "agent commander" or orchestrator.
Instead of writing every line of code or manually configuring every cloud resource, engineers will increasingly define the goals, set the parameters, and review the output of AI agents. Their expertise will shift towards designing robust systems, understanding complex interactions, and ensuring the AI systems align with business objectives.
Pulumi won't be alone in this market. Competitors like Harness, which calls it the "agentic SDLC," and GitHub, which pitched "agentic software delivery" at its Universe conference, are also exploring this space, according to GeekWire. This indicates a broader industry trend towards greater automation in software operations, pushing businesses to rethink their team structures and skill requirements.
The Bigger Picture
Pulumi's "absurd ambition" points to a future where the bottleneck in software development isn't the speed of human coding, but the scope of our imagination. As AI agents become more sophisticated, the focus shifts from manual execution to strategic direction.
For business leaders, this means preparing your teams for an evolution, not an elimination, of roles. The engineers who thrive will be those who can design, manage, and audit AI-driven systems, transforming from individual contributors to orchestrators of intelligent automation.
The question isn't whether AI will change software delivery, but how quickly businesses can adapt their talent and processes to harness its full potential. The era of the "agent commander" is dawning, demanding a new kind of leadership in technology.
Are you ready to command your own fleet of AI agents?
Quick Answers
What are Pulumi Delivery and Pulumi Agents?
Pulumi Delivery automates the software release process, from testing to rollout, allowing AI agents to code release steps and assess risks. Pulumi Agents are designed for operational tasks like investigating outages, reducing cloud costs, and resolving security issues within defined limits.
How do Pulumi's AI agents ensure safety and human oversight?
The AI systems incorporate 'decision gates' where an AI model like Jev assesses risks, stopping releases if thresholds are exceeded. Additionally, all changes proposed by AI agents appear as code for human teams to review and approve, maintaining transparency and control.
What does Pulumi's AI strategy mean for the role of engineers?
The role of engineers is evolving from primary coders to 'agent commanders' or orchestrators. They will focus on defining goals, setting parameters for AI agents, and reviewing AI-generated output, enabling them to manage more infrastructure and focus on higher-value, strategic tasks.
Sources
- GeekWire on MSN18 Nov · 14:27 IST
Facts are credited to the outlets above. The analysis, calculations and Signal Check are by the Finnect AI Insights desk.