In a recent TFiR interview, Swapnil Bhartiya spoke with Waldemar Hummer, Co-founder and CTO at LocalStack, about the operational friction plaguing cloud development workflows, the strategic shift toward local AWS emulation, and how LocalStack is becoming the de facto AI sandbox for validating agent-generated code.

The Cloud Development Friction Problem

Hummer identified slow feedback loops as the primary barrier to velocity in modern cloud development. Traditional AWS testing requires pushing every code change to remote infrastructure, creating deployment cycles measured in minutes or hours instead of seconds.

Q: What problem is LocalStack solving for cloud developers?

Waldemar Hummer: “When you develop against the cloud, the developer experience is often subpar, not ideal, and not optimal. What we do with LocalStack is give you the ability to have a much more snappy experience on the local machine. You’re talking not about minutes or even hours to deploy, but rather seconds for the feedback cycle.”

He noted that this friction is particularly severe in regulated industries like banking and insurance, where developers often lack direct cloud access entirely. In these environments, every code change must flow through CI/CD pipelines that can take 30 minutes per deployment, creating a compounding tax on iteration speed.

Q: How does this play out in regulated industries?

Waldemar Hummer: “This problem is even more aggravated by some regulated industries, where sometimes developers don’t even have access to the cloud, so they might need to push changes through some CI/CD pipeline that takes half an hour to deploy. The feedback cycles become even slower and even more tedious.”

LocalStack as an AI Sandbox for Agentic Workflows

With the rise of AI-generated code, Hummer emphasized that LocalStack’s value proposition extends beyond traditional development workflows. The platform now functions as a critical sandbox for validating code produced by agentic AI systems—a use case that addresses growing concerns around “AI slop” and untrusted agent outputs.

Q: How does agentic AI change the testing equation?

Waldemar Hummer: “The whole world of software is changing. With agentic AI, this even increases the necessity to test and validate code very quickly, because you cannot always trust the code that’s being generated by these agents. You need a way to test in an efficient and also sandboxed manner, without risking an agent pushing changes to the real cloud environment. That’s where our AI sandbox use case comes in.”

LocalStack is actively partnering with agentic AI platforms to embed local cloud emulation directly into agent workflows, enabling transparent validation without developers needing to configure sandbox environments manually.

Q: What does the LocalStack integration with agentic platforms look like?

Waldemar Hummer: “We’re starting conversations to really embed LocalStack more deeply as an AI sandbox in some of these agentic coding platforms, making it a seamless experience. Whenever you have a use case for developing your cloud app, you can just come to some of these platforms and have the environment almost transparently available, without even having to think that LocalStack is running in the background.”

AWS Partnership and Multi-Cloud Expansion

Despite initial perceptions that LocalStack might compete with AWS, Hummer described a deepening partnership with Amazon’s product teams. The collaboration includes same-day feature launches, native integrations in AWS developer tooling, and co-development initiatives with the AWS Lambda team.

Q: How does AWS view LocalStack?

Waldemar Hummer: “We’ve started partnering very closely with AWS. We partner with their product teams and sometimes have co-launches, like same-day launches last year at AWS re:Invent. We launched some features with the AWS Lambda team because there’s a lot of demand from their customer base for enabling a local developer environment. Last year, we also rolled out a feature for a Visual Studio Code integration, so the AWS Toolkit for Visual Studio Code now has a native integration for LocalStack.”

LocalStack is also expanding beyond AWS into multi-cloud territory. The platform launched a Snowflake emulator in 2024 and is preparing to release Azure emulation capabilities, positioning itself as a unified local development platform across hyperscalers.

Q: What’s the multi-cloud strategy?

Waldemar Hummer: “We’ve been focusing on AWS as the primary flagship product. But more recently, we put out a Snowflake emulator about a year ago. And we’re also now venturing into Azure as the next cloud provider. What we’re looking to establish is a multi-cloud platform that allows you to test end-to-end your applications.”

Chaos Engineering and Resilience Testing

Beyond basic emulation, LocalStack introduced chaos engineering capabilities that allow developers to inject faults, latencies, and failure scenarios into local cloud environments. This enables proactive resilience testing before applications reach production.

Q: What chaos engineering features does LocalStack offer?

Waldemar Hummer: “We recently introduced a feature called chaos engineering. Chaos engineering allows you to inject errors or latency into the APIs to test your applications for resiliency. What happens if us-east-1 goes down? What happens if my database is at capacity? We can simulate these situations in LocalStack and make sure that your application is resilient to these kinds of faults, because things do happen in the cloud. Just a couple of months ago, us-east-1 experienced outages, and it’s good to be prepared for these error scenarios beforehand.”

Platform Engineering Adoption Model

Hummer outlined LocalStack’s dual go-to-market motion: bottom-up developer adoption through self-serve downloads, and enterprise platform engineering teams implementing the tool as organization-wide infrastructure.

Q: How do platform engineering teams use LocalStack?

Waldemar Hummer: “We work with platform engineering teams that are building the developer experience for the organization more broadly. Those are typically our main inspiration and partners in introducing LocalStack as a key value driver across the organization. We can adjust and tailor it to their specific environments. There is a turnkey aspect to it, but we also do a lot of customization with the enterprises that we work with.”

LocalStack offers features specifically designed for platform engineering use cases, including Cloud Pods—persistent snapshots of local cloud environments that can be shared across teams for reproducible testing.

Q: What are Cloud Pods?

Waldemar Hummer: “We have a feature that we call Cloud Pods. It’s basically like taking a persistent snapshot of your container, and you can easily share that with your team members. It’s almost the equivalent of taking a full snapshot of an AWS account and then moving that somewhere else, which would be difficult to achieve in the real cloud. If you have a failing test in your CI pipeline, you can just persist the state, pull down the snapshot, and reproduce everything locally, so you have full insight and debuggability.”

Sovereign Cloud and European Infrastructure Implications

Hummer noted growing interest in sovereign cloud initiatives across Europe, driven by regulatory requirements and geopolitical considerations. LocalStack’s local-first architecture aligns naturally with sovereignty concerns, as workloads execute entirely on-premises without reliance on US-based hyperscaler infrastructure.

Q: How does LocalStack fit into sovereign cloud strategies?

Waldemar Hummer: “We’re having a lot of conversations at KubeCon here about sovereign cloud, and it’s a trend that’s gaining momentum. In general, we’re seeing a growing trend toward decentralization. Europe is very much at the forefront of this. With our solution, we give you the ability to run your workloads literally on your machine, on your laptop. This is very much to our advantage.”

Open Source Foundations and Community Strategy

LocalStack’s 400 million Docker pulls and 65,000 GitHub stars reflect its roots in open source community adoption. The company recently consolidated its Community Edition and Enterprise offerings into a unified image requiring user registration, aiming to improve developer experience while maintaining direct customer engagement.

Q: What’s the open source strategy?

Waldemar Hummer: “We definitely have a strong DNA in open source. We contribute part of our internal platform to open source projects. The love from the community has helped us tremendously in making the product widely known and widely adopted. Now we’re increasingly also targeting enterprise buyer personas, but we’re still deeply rooted in the developer community with our product.”

Q: Why did you unify the Community and Enterprise images?

Waldemar Hummer: “We recently made a change to combine the images into one. We want to provide a more unified developer experience because we saw that there was a bit of fragmentation between the OSS version and our advanced offerings. The main change is that we now require users to sign up. You provide your email address, create an account, and then you can leverage the full power of the LocalStack cloud platform with a web application and DevEx features. That allows us to interact more directly with our users and optimize the product experience over time.”

The Future of Cloud Development: Decentralization and Quality Gates

Hummer predicted a fundamental shift in software development driven by agentic AI and a broader trend toward decentralization. He argued that traditional development processes—code reviews, feature planning, open source governance—must adapt to a reality where agents generate code at unprecedented velocity.

Q: How will agentic AI reshape software development?

Waldemar Hummer: “We need to fundamentally rethink how software is being built, because a lot of processes have been built around humans—the code review, open source, feature planning, prioritization. All of these basic principles of engineering and product development need to shift and adapt to this new reality. We need to be much more diligent on being able to trust the code that’s generated by agents, have very strict quality gates, and make sure that we have a very tight testing and quality assurance loop.”

He emphasized that every AI model released today represents the worst version of that technology moving forward, reinforcing the urgency for robust testing infrastructure as agent capabilities accelerate.

Q: What’s the trajectory for AI-generated code quality?

Waldemar Hummer: “Any model that we see is the worst version of that model for now and into the future. It’s only going to get better from here. If we’re already quite excited and impressed with Claude and some of the other models, then it’s only going to improve. This aligns closely with our storyline of testing frequently and early—quality assurance is important. With that comes a certain efficiency that you want to give to these models. They need a sandbox that they can test against.”