Distributed application platform provider Akka said today it’s launching two new deployment options for developers looking for more versatile ways to deploy agent artificial intelligence applications at large scale.

The company, which rebranded last year as Akka but is still officially known as Lightbend Inc., said developers now have the choice to self-host their applications in two places – either on Akka itself or across multiple cloud regions.

By enabling companies to self-host their agentic AI applications, developers no longer have to commit to using the full Akka Platform, the company said.

The Akka development platform first became popular with developers around five years ago, providing a versatile platform and toolkit for building concurrent, distributed, scalable and resilient applications on the Java Virtual Machine. Its popularity stems from its efficient use of the Actor Model, which enables lightweight, independent units of code, known as “actors” to communicate asynchronously via messages, simplifying concurrent and distributed programming.

It has become a popular choice for building cloud-native services, microservices, streaming pipelines and internet of things applications because it overcomes many of the challenges associated with traditional object-oriented programming. For instance, it eliminates the need for mechanisms such as synchronization locks and semaphores, which are required to mitigate problems associated with the use of shared memory

More recently, Akka has become a popular choice for developers building agentic applications, which utilize AI agents to perform business tasks on behalf of users with minimal supervision. There are significant benefits to building agentic AI systems on a conversation-centered platform like Akka. While traditional software-as-a-service apps are built on stateless business logic, executing operations against relational databases, agentic services maintain their own state, storing each state to keep track of how they reached their current state.

This is important as it eliminates the unpredictable behavior associated with agentic AI systems built on stateless business logic.

But previously, the Akka SDK required developers to use the Akka Platform, which meant they had no control over where their apps were deployed. That changes with Akka’s new deployment options.

Now, developers have two new options, including the ability to deploy on self-managed Akka nodes built using the Akka SDK on any cloud infrastructure. The latest version of the Akka SDK features a self-managed build option for users to create services that can execute without the Akka platform. Binaries are packaged as Docker images that can be deployed on any container platform-as-a-service, the company said, as well as Kubernetes platforms, cloud-based virtual machines, bare metal hardware and edge nodes.

In addition, teams also have the option to run their own Akka Platform region without any dependency on Akka’s control planes, the company said. By choosing this option, teams will need to install, maintain and manage the control plane by themselves, as opposed to the fully managed experience on the main Akka Platform.

The company said the new deployment options are a necessary evolution of its application stack, giving developers the freedom they need to maintain full control over their agentic AI applications, building them on the most appropriate infrastructure and location.

Akka Chief Executive Tyler Jewell told SiliconANGLE that Akka’s platform has a reputation as being the first microservices framework that enables almost infinite scalability for distributed applications and the ability to quickly recover from any failure.

“Now, Akka enables you to build distributed services that can deploy into any existing infrastructure whether it’s bare metal, VMs, containers, container PaaS, or Kubernetes,” he said. “Akka services cluster-from-within, enabling them to manage their own elasticity, resilience and agility without the complexity that comes from standing up a platform.”

According to Jewell, agentic AI has become a top priority for enterprises in many industries, as it has the potential to completely replace existing enterprise software as we know it today.

“We’re making it easy for enterprises to build their distributed systems, including agentic AI deployments, without having to commit to Akka’s Platform,” he added.

Image: SiliconANGLE/Dreamina