Pulumi is in New York June 17 and 18, with three ways to get into deploying AI agents: a hallway chat at AWS Summit, a hands-on workshop, and the New York Pulumi User Group kickoff. Workshop seats are limited: https://www.aicamp.ai/event/eventdetails/W2026061714

Heading to DevOpsCon Berlin June 16-17. Visit our Silver sponsor booth or catch Wladi Mitzel's talk "Designed for agents, useful for humans: Five IaC practices that pay off twice." Grounded in Pulumi telemetry across ~10M production deployments. Practices apply to Pulumi, Terraform, and OpenTofu.

Pulumi Deployments can now trigger on git tag pushes. Push a version tag like v1.2.0 and Pulumi runs pulumi up automatically — with glob-based tag filters so you only deploy the releases you choose.

Cloud AI APIs keep agents easy until tradeoffs show up: data leaves your network, offline work breaks, and every token has a meter. Pablo Seibelt rebuilt the stack locally with Gemma 4, llama.cpp, k3d, Pulumi, and Tailscale. The interesting part is what runs where, and why.

Build isolated EKS test environments with Pulumi and vCluster. Model the host cluster, vCluster tenants, guardrails, and cleanup as code.

2x is mostly prompting. 10x is mostly plumbing. The interesting question: what changes when the five agents are touching infrastructure?

Why choose Pulumi over Terraform? Practical examples for refactoring, secrets, testing, provider wiring, safer changes, and honest IaC caveats.

Pulumi and Google are hosting a fireside in SF: IaC in the Age of AI and Agents, with Jay Smith, Staff Customer Engineer at Google Cloud. Candid takes, real war stories, drinks on us, and a room full of platform engineers and SREs. The Gambit, June 3 at 6 PM. RSVP below.

Pulumi is the Platinum sponsor of CascadiaJS 2026, June 1–2 at Town Hall Seattle. Joe Duffy keynote: The Last Mile Is Code. Workshop: deploy AI agents to AWS with Pulumi + Bedrock AgentCore. Register with our discount.

Your Claude Code session plateaus in a real codebase because the layer around the agent is empty, not because the model is weak. Seven pieces of harness, in order of where to start. www.pulumi.com/blog/stop-tu...

Every PR review opens with the same question: what does this change actually do? Mention @pulumi-neo and you get the answer in the thread. Mention @Neo in Slack and you get the same agent there too, with the same context and permissions it has anywhere else.

Recurring platform work slips. Provider versions fall behind, drift accumulates between checks, and the quarterly audit keeps getting pushed back. Pulumi Neo can now run that work for you and ship each result as a PR.

Five engineers. Nineteen clients. Custom code shipped within a day of signing. Ewan Dawson's seven rules for an AI-native software factory — where agents write nearly all the code.

Infrastructure work often lives between browser tabs: PagerDuty in one, Datadog in another, the AWS console in a third, Pulumi Cloud in a fourth. With today's release, Neo pulls it all into one conversation.

Our CLI traffic shifted: people and agents running commands in the same session. We rebuilt the Pulumi CLI for both — guessable commands, all of Pulumi Cloud reachable from the terminal, output that's readable to either. Here's what changed.

Last fall we wrote up 10 things you could do with Pulumi Neo. Since then teams have handed Neo real work: drift, dependency upgrades, on-call triage, CIS Benchmark fixes that show up while you sleep. Here are 10 more.

We’re thrilled to welcome Joe Duffy, co-founder and CEO of @pulumi.com, as a keynote speaker at CascadiaJS! From programming languages to cloud infrastructure, Joe has spent his career building the tools developers use every day. See him at CascadiaJS, the web + AI conference for devs, by devs.

I'm happy to start another year as a @pulumi.com Puluminary! I always love and appreciate the swag! Also, bonus points for having some of the coolest shipping boxes I've seen.

A year ago, building an agent meant a framework, a RAG pipeline, and a pile of glue code. Today most of that middle layer is gone. Here is what replaced it.

Stripe ships over a thousand AI-authored PRs a week. The pattern behind it has a name: the dark factory. The infrastructure factory is different. Here's what happens when the factory floor is your Pulumi state graph.

DevOpsDays + AI Nashville is coming up next week (May 14–15), and we've got five free passes to give away! If you're in the area and want to be there for some amazing talks and open spaces, just follow the link, no strings. See you there! https://info.pulumi.com/pulumi-devops-days-nashville

96% of enterprises run AI agents in production. 12% have any central way to manage them. Developers are shipping agents faster than platforms can govern. The governance layer you need is already in your stack. Here's where. https://www.pulumi.com/blog/agent-sprawl-iac-platform-is-the-answer/

We're packing up the swag bins and heading to Google Cloud Next '26 next week in Las Vegas, April 22–24. 🛫 Come find us at Booth 3624 to talk Google Cloud infrastructure, IaC, and all things Pulumi, and while you're there, grab yourself a plushie. See you there! 👋 www.pulumi.com/google-cloud...

Headed to Google Cloud Next? Join us for a breakout session and demo with Google Cloud and TelevisaUnivision about getting containers into production: From Code to Production: The Path of Least Resistance Day 2, April 23 @ 12:00 - 12:45 PM www.googlecloudevents.com/next-vegas/s... See you there!

Most orgs have experimented with AI workloads on Kubernetes, but very few run them in production daily. That gap defined KubeCon EU 2026. llm-d, SPIFFE-based agent identity, inference gateways. The pieces are forming fast.

We eliminated all long-lived CI secrets across 70+ repos using Pulumi ESC and OIDC. Short-lived credentials mean a compromised GitHub Action gets nothing persistent. Here's how we did it:

Open source makes the world go around! In honor of FOSS month, our SIGs wanted to thank the projects we're relying on most. There are a ton, so please join us as we celebrate them all! 🙌

Engineers waste deploy cycles on config typos and invalid values. Pulumi ESC’s new fn::validate enforces JSON Schema at save time so misconfigurations fail early, not during a deploy. Define your rules once and catch errors before they hit stacks. www.pulumi.com/blog/esc-sch...

We let an LLM generate SQL for analytics. The answers looked right and were wrong. Missing joins. Bad filters. Wrong assumptions. Our data team shares what failed and what worked instead 👇 www.pulumi.com/blog/how-we-...

Does AI not remember your project rules? Stop repeating yourself! Pulumi Neo now supports AGENTS.md so conventions like naming, structure, and required steps live in the repo and are reused automatically. www.pulumi.com/blog/pulumi-...

We heard you! The Pulumi Cloud REST API now has an OpenAPI 3.0 spec! Fetch the spec, generate clients, keep everything in sync. www.pulumi.com/blog/announc...

Neo now supports task sharing. Share any task with your team: the original prompt, Neo's reasoning, the actions it took, and the outcome. Full context preserved. www.pulumi.com/blog/neo-tas...

Pulumi Agent Skills are structured knowledge packages for AI coding assistants. Install them once and get correct, idiomatic Pulumi guidance inside your preferred tool, including Claude Code, GitHub Copilot, Cursor, VS Code, Codex, and Gemini CLI. www.pulumi.com/blog/pulumi-...

How do you run an AI assistant without exposing it to the internet? This walkthrough shows @openclaw-x.bsky.social on AWS or Hetzner using Pulumi and @tailscale.com for private access, reproducible infrastructure, and clean lifecycle control. www.pulumi.com/blog/deploy-...

What happens when AI isn’t babysat and infrastructure is written in familiar programming languages? This experiment using the Ralph Wiggum loop shows Claude building and deploying a serverless SaaS on AWS with Pulumi. www.pulumi.com/blog/how-ral...

Some infrastructure needs to be replaced on purpose. Pulumi now lets you control when resources are recreated using the 'replacementTrigger' resource option, without manual flags or brittle workarounds. www.pulumi.com/blog/trigger...

We'll be at @cfgmgmtcamp.bsky.social in Ghent! If you’ll be there, stop by the Pulumi booth for a demo, meet the team, and learn more about how Pulumi fits into the config-management picture, helping drive tools like Ansible and manage complex Kubernetes deployments without wrestling with YAML.

Migrating from CDK, Terraform, or Azure ARM usually means risk, downtime, or both. Neo changes that. It uses your existing IaC state to move infrastructure to Pulumi with zero downtime and a verified zero change preview. www.pulumi.com/blog/neo-mig...

Running Pulumi Cloud self-hosted means stricter security and no room for shared credentials. Pulumi IAM now supports self-hosted deployments with scoped tokens, custom roles, and least-privilege access for teams and automation. www.pulumi.com/blog/pulumi-...

The new Pulumi ESC Web Editor is live. Switch between YAML and a rich UI to manage secrets, providers, and exports with full visibility into your environment. It makes setting up OIDC, editing secrets, and sharing config much easier. Learn more at www.pulumi.com/blog/new-esc...

Ever need a value to stick around between pulumi up runs? Meet Stash, a new built-in resource to Pulumi IaC for persisting data in your stack's state. www.pulumi.com/blog/introdu...

Ingress NGINX is retiring in 2026. Kubernetes teams should start planning now. The Gateway API offers a standard, expressive model for traffic management, and kgateway provides a production-ready path for migration with lower risk. www.pulumi.com/blog/ingress...

Moving AI agents from local prototypes to production is where most teams struggle. This walkthrough shows how to deploy agents with Amazon Bedrock AgentCore using real infrastructure patterns for security, identity, and runtime management. www.pulumi.com/blog/from-wo...

CDKTF is deprecated. Teams using TypeScript or Python for infrastructure now need a path forward. Options range from returning to HCL to moving to a language-first IaC model that can coexist with Terraform. Follow along 👇 www.pulumi.com/blog/cdktf-i...

Your GPUs might be burning money right now without you knowing. Idle clusters and forgotten experiments add up fast. What if your infrastructure warned you before the cost hit your bill? www.pulumi.com/product/supe...

Learn how to deploy a multi service application on Azure Container Apps in a live workshop on January 15. See how to structure services, handle configuration, and ship a real cloud application on Azure. Register here: www.bigmarker.com/pulumi/build...

AWS re:Invent 2025 introduced major shifts in AI and cloud infrastructure. Nova Forge, Trainium 3, and AgentCore point toward more integrated, automated systems. Pulumi’s roundup breaks down what these launches mean for engineers: www.pulumi.com/blog/aws-rei...

Bring Cursor into your cloud workflow with the Pulumi Remote MCP Server. Let Cursor inspect stacks, flag drift or issues, and pass complex updates to Pulumi Neo for automated execution. Ready to try it? 👉 www.pulumi.com/blog/remote-...

Pulumi Kubernetes Operator + Kargo turn infra-as-code updates into a real promotion pipeline—staged rollouts, approvals, verification, and audit history without “deploy everywhere at once” chaos. 👉 www.pulumi.com/blog/pulumi-...

The cloud you know is changing fast. AI, IaC, Kubernetes, and platform engineering are rewriting how teams build and operate infrastructure. Don’t fall behind. See the 10 trends shaping 2026 and beyond at www.pulumi.com/blog/future-...