April 2026 Summaries

Temporal is intensifying its commitment to artificial intelligence by integrating key talent from Chkk and Adviser Labs, aiming to enhance its capabilities in building durable and reliable AI systems. Chkk's founder, Awais Nemat, will spearhead the engineering efforts, bringing extensive experience from AWS and experience with Temporal's platform. Adviser Labs contributes fundamental AI research from the Stanford AI Lab and expertise in AI-powered cloud orchestration. This strategic move aims to address reliability challenges in AI through initiatives focused on agentic AI use cases, AI-powered developer tools, and collaborations with major tech entities like OpenAI and Google ADK. Temporal's goal is to create a robust foundation for AI agent communication and to lower barriers for deploying agents, aligning with its vision to become an AI-native company. The upcoming Replay event will further explore these advancements and the future trajectory of AI within the company.

Layer.ai has evolved from a tool for generating individual game assets to a comprehensive creative operating system for mobile game studios, leveraging Temporal to manage complex production pipelines efficiently. Initially, the platform handled almost all asset generation through simple inference forms; however, as demand increased, Layer transitioned to a more sophisticated node-based workflow system, allowing for the parallel generation of hundreds or thousands of assets. This shift was partly driven by the need to support video production, a more resource-intensive process than image generation, as video game studios increasingly use short-form videos for marketing on platforms like TikTok and YouTube. Layer has also introduced an MCP connector, enabling LLM agents to autonomously invoke workflows, thereby eliminating human bottlenecks in the creative process and significantly reducing production costs for formats such as playable ads. The company runs a self-hosted Temporal cluster to maintain control and ensure reliability, implementing strategies like task queue isolation and proactive monitoring to manage workloads effectively. This technical evolution reflects an ongoing commitment to scalability and innovation, as Layer continues to enhance its infrastructure to support growing orchestration complexities.

In this newsletter, Temporal shares updates from April, including new product releases and integrations that enhance its capabilities for workflow management and distributed systems. Key highlights include the OpenAI Agents SDK sandbox integration, Google ADK integration, and the availability of Multi-cloud and Multi-region Replication for improved failover capabilities. Additionally, the OpenMetrics Endpoint for Temporal Cloud is generally available, allowing users to export metrics for better observability, while the Public Preview of Billable Actions Metrics offers insights into Temporal Cloud Actions. The newsletter also introduces Temporal's External Storage feature in pre-release and announces that the Temporal Plugin on Cursor Marketplace is live. Exciting developments include the GA release of the Python SDK and public previews for TypeScript and .NET SDKs. The newsletter features a builder spotlight on XY, an AI-native healthcare company using Temporal to streamline complex workflows, and provides insights into community projects and challenges, such as GrowthX's output.ai and troubleshooting efforts by community members. The newsletter encourages readers to join upcoming live events and provides learning resources for understanding Temporal's design patterns.

The text explores the parallels between past technological transitions, such as the move to cloud-native environments and microservices architectures, and current developments in the AI space, particularly with AI agents. It highlights how many core challenges, like managing complexity, orchestrating systems, and handling state and time, have persisted across these transitions. The text discusses the evolution from monolithic systems to microservices and now to microagents in AI, emphasizing the benefits of breaking down large systems into smaller, more manageable components. It also touches on the shift from request/response models to event-driven architectures, as well as the importance of effectively managing short-term memory and state in distributed systems. The narrative underscores the value of applying lessons learned from previous experiences in microservices to the current AI landscape, advocating for a thoughtful approach to leveraging past insights while adapting to new capacities offered by AI technologies.

Temporal is a platform designed to enhance the reliability and efficiency of AI agent systems by addressing challenges in orchestration, versioning, and cost management. Unlike traditional agentic AI systems that often face issues such as timeouts, conflicting changes, and escalating costs, Temporal provides a robust architecture that ensures agents operate within guardrails, maintain data security, and manage long-running processes effectively. It offers solutions for different types of agents, including conversational, event stream processors, ambient, and multi-agent systems, by providing features like workflow versioning, retry policies, and context preparation. Temporal's orchestration capabilities, such as Temporal Nexus, allow for seamless coordination among multiple teams building diverse agents, transforming agent interactions into secure, observable, and versioned calls. This structured approach not only ensures scalability and durability but also provides visibility into agent operations, enabling better debugging, auditing, and analytics, ultimately making AI agent systems more trustworthy and manageable for businesses.

XY has developed an AI agent orchestration platform specifically designed to automate complex healthcare workflows, utilizing the Temporal framework to manage the intricacies of insurance claims processing. The platform addresses the challenges inherent in healthcare automation, such as multi-system interactions, human-in-the-loop processes, data sensitivity, and the need for robust reliability and scalability. Traditional orchestration tools fall short in this domain due to healthcare's unpredictable workflows. XY's solution employs a YAML-based Domain Specific Language (DSL) built on Temporal, enabling a single, generic workflow class to manage diverse tasks, while maintaining a separation between workflow logic and business logic. The platform's AI-powered Planner Agent translates natural language requests into executable YAML, streamlining workflow creation and ensuring compliance with security and data handling standards. Temporal's architecture supports efficient scaling and versioning, allowing for seamless evolution of the DSL without disrupting active workflows, which is crucial in the dynamic healthcare environment. This setup enables XY to handle large-scale operations efficiently, while ensuring cost-effectiveness by aligning infrastructure usage with actual workload demands.

The text explores the use of Temporal's Worker Versioning on Kubernetes to manage workflow deployments efficiently while avoiding non-determinism errors that can occur when code changes are made during in-flight workflows. It details a process of building a research agent workflow, intentionally breaking it with naive deployments to demonstrate the problem, and then resolving it through Temporal's Worker Versioning. The approach ensures that workflows can safely roll out code changes without losing in-flight work or experiencing deployment anxiety by using multiple versions of workers running simultaneously on the same task queue. The Temporal Worker Controller on Kubernetes automates the management of these versions, allowing old workflows to complete on their original version while new ones start on updated workers, preventing conflicts and preserving computational resources. The text also delves into handling long-running workflows, migrating them with Continue-as-New and AUTO_UPGRADE for seamless transitions, and managing idle workflows with wake-up signals to detect version changes. The broader context of scaling and rolling out deployments progressively for high availability is also covered, highlighting the strategic use of Temporal's features for robust workflow management.

Kubernetes has emerged as the standard platform for deploying scalable production software, and the recently released 1.0.0 version of Temporal Service's Helm Charts marks a significant milestone, offering stability without expected breaking changes in future releases. Originating as a side project by Temporal engineers, these Helm Charts have evolved through robust community collaboration, with contributions from nearly 150 individuals, most of whom are not Temporal employees. The new release introduces changes that require users to manually provide databases and optional components like Elasticsearch and Prometheus, as the Helm Charts no longer include sub-charts for these. Configuration updates are necessary due to changes in the values.yaml file structure, and new features have been added, including configurable readiness probes, customizable deployment strategies, and enhanced security measures. For users navigating migration or starting anew, resources such as the UPGRADING.md file, the README, and the #helm-charts channel in Temporal’s Community Slack offer guidance, while GitHub remains the platform for reporting bugs and continuing community collaboration.

Kyndryl's initiative to integrate Google's Agent Development Kit (ADK) with Temporal highlighted the demand for such integrations, prompting Temporal to officially support this functionality. The integration aims to enhance the agentic loop's durability by running LLM and tool calls as individual Activities within Temporal Workflows, allowing precise retries and maintaining visibility into each step. This approach resolves limitations of previous harness patterns that lacked detailed insight and required extensive manual recovery efforts. The integration, which initially supports Python, was simplified by creating workflow-safe versions of ADK that separate deterministic and non-deterministic processes, ensuring seamless serialization across process boundaries. The collaboration with Google led to the introduction of an abstraction layer for time and UUIDs, maintaining determinism in Workflow executions. Temporal's AI Partner Program, leveraging insights from the ADK integration, invites teams to collaborate on building AI frameworks and tools, while developers benefit from an enhanced debugging experience with real-time visibility of agent activities in multi-agent setups.

OpenAI recently introduced sandbox support within its Agents SDK, enabling agents to execute code, manipulate files, and run shell commands in isolated environments using providers like Modal, Daytona, Docker, and E2B. This new feature, developed with input from OpenAI's engineering team, includes a Temporal extension that allows sandbox agents to function as durable Temporal workflows, maintaining state and reducing resource waste by idling without consuming compute resources. The demo provided showcases the architecture consisting of AgentWorkflow, SessionManagerWorkflow, and a Textual-based terminal interface (TUI), highlighting capabilities such as session forking and backend switching while preserving workspace state. This advancement ensures agents are more reliable and production-ready through enhanced fault tolerance and state management, paving the way for scaling to thousands of concurrent sessions without idle sandbox costs. The integration with Temporal provides durability and persists the workflow state across restarts, with plans for further integrations with platforms like Pydantic AI and Amazon Bedrock Strands.

Samar Abbas, co-founder and CEO of Temporal, draws an analogy between the current state of AI agents and the early MS-DOS era of computing, emphasizing the democratization and potential of AI akin to the evolution of personal computing. Reflecting on his own journey from programming in Pakistan to building distributed systems for major tech companies, Abbas discusses the parallels between early computing architectures and modern AI systems, noting the need for robust sandboxes and distributed frameworks as AI evolves. He highlights the security challenges inherent in deploying AI agents, citing the risks of running sensitive data on unsecured machines, and suggests a future shift towards centralized, secure environments to manage these agents. Abbas envisions a future where AI agents will work alongside human employees, resulting in a dynamic, elastic workforce and necessitating new tools and frameworks to address the complexities of distributed systems and state management. The conversation underscores the transformative impact of AI on organizational structures and the pressing need to develop novel system primitives to accommodate the massive scale of future AI applications.

Replay '26, taking place at Moscone South in San Francisco from May 5–7, is a conference designed for engineers facing diverse challenges in their work environments. The event caters to various professional archetypes, such as those modernizing outdated systems, leading AI-driven projects, ensuring system reliability, or designing scalable architectures. Attendees can expect to engage in sessions that cover critical topics like system migrations, AI orchestration, observability, and platform engineering. Beyond structured talks, the conference emphasizes the value of community and shared experiences, offering opportunities for informal discussions and networking with peers who understand the complexities of engineering challenges. With a focus on practical solutions and community building, Replay '26 promises to be a valuable gathering for engineers seeking insight and camaraderie in their respective fields.

Temporal Workers operate differently from standard web services as they long-poll the Temporal Server for tasks without receiving inbound HTTP traffic, eliminating the need for a load balancer. Amazon ECS with Fargate is well-suited for deploying these Workers, which are long-running processes that execute workflows and activities locally. The guide details the necessary ECS configuration, including setting up health checks, handling graceful shutdowns, and managing infrastructure with IAM roles and task definitions. Temporal Workers are ideal for Fargate Spot due to their stateless nature, offering significant cost savings as they retry in-progress activities on another Worker if a Spot instance is reclaimed. Scaling these Workers requires different considerations than web services, focusing on CPU utilization and asymmetric cooldowns, while observability is achieved through CloudWatch logs, Container Insights, and Temporal Cloud metrics. The guide suggests starting with ECS for those new to Kubernetes and considering EKS for larger fleets or when specific Kubernetes functionalities are needed, without changing the Worker code but only the surrounding infrastructure.

AI agents are increasingly capable of performing complex tasks autonomously, but their reliability remains a significant challenge, especially during extended workflows. This issue stems not from the intelligence of AI models, which have continued to improve in accuracy and trustworthiness, but from the lack of robust infrastructure that can handle failures mid-process. An incident involving Google's Antigravity AI coding assistant highlighted this vulnerability when it mistakenly erased a user's entire D: drive, illustrating the urgent need for systems that can recover from partial failures without starting over. The AI industry has traditionally focused on enhancing model accuracy, but as AI agents transition from suggestion-based systems to action-oriented systems, the need for durable execution infrastructure becomes crucial. This infrastructure should function as a digital bookmark to allow seamless resumption of tasks, addressing the compound failure problem that arises when AI systems manage long-running operations autonomously. The integration of such infrastructure, akin to what companies like Temporal provide, is essential to ensure AI agents can reliably execute tasks over extended periods, preventing irreversible actions and maintaining continuity.