Durable Digest: August 2026

Highlights this month include new ways to run Workers without managing infrastructure, organize Temporal Cloud resources, and build more durable AI applications. You’ll also find new controls for managing running Workflows and Activities, expert-reviewed Validated Patterns for production architectures, answers to real questions from the Temporal community, and opportunities to connect with the team at upcoming events.

See how Temporal is making it easier to build and operate reliable applications, from distributed systems to AI agents. Explore what’s new, learn from the Temporal community, or start building for free on Cloud.

Recently shipped#

State of Development Report 2026: AI agent use is up 71% year over year, but engineering teams still face challenges around state, debugging, and reliability. See how engineers and engineering leaders are using AI agents, what’s working, and where they’re running into challenges.
Serverless Workers for AWS Lambda is available in Public Preview. Serverless Workers for Lambda offers zero infrastructure management, instant autoscaling, scale to zero, built-in observability, and simplified versioning by default. Temporal Cloud customers simply upload their Worker code to a Lambda function, provide Temporal with the ARN and an IAM role, and Temporal handles function invocation automatically. When there’s no work, functions scale to zero, and no billing is incurred.
Projects for Temporal Cloud are now in Pre-Release. Projects give teams a first-class way to organize Temporal Cloud resources, Namespaces, Nexus Endpoints, and more around how the organization actually operates, whether by team, environment, service, or cost center. We provide six predefined Project-level roles so platform teams can delegate scoped access without broad Account-level permissions.
Standalone Nexus Operations are now in Pre-Release. Standalone Nexus Operations let you invoke a Nexus Operation directly from the Temporal SDK Client, CLI, or an Activity, with no proxy Workflow or bespoke gateway required. They provide Durable Execution guarantees, including at-least-once delivery, retries, deduplication, and visibility, across five SDKs.
The Google GenAI integration is now in Public Preview. This plugin lets developers write against the Google Gen AI SDK and get Temporal’s durability automatically, so you no longer have to hand-wrap every Gemini model and tool call in your own Activities.
Our Deep Agents integration is now in Pre-Release. This plugin wraps every LangChain Deep Agents model and tool call in a Temporal Activity, so your agent loop runs inside a deterministic, replayable Workflow and resumes exactly where it left off after a crash, with no changes to your Deep Agents code required.
Activity Operations are now available in Public Preview. Activity Operations give operators surgical control over individual misbehaving Activities. Operators can Pause/Unpause, Reset, and Update Options, including timeouts, Retry Policies, and Task Queues, without disrupting the parent Workflow. Batch support is available through visibility queries.
Workflow Pause is now in Pre-Release. Workflow Pause lets you pause a running Workflow entirely. Pausing creates no new Workflow Tasks or Activity dispatches, and Signals remain queued until the Workflow is unpaused. This gives operators time to investigate a broken Workflow without losing state, fix the root cause, and reset the Workflow instead of terminating it.
Worker Performance is now Generally Available. Poller Autoscaling dynamically tunes your Worker’s poller count, so you no longer have to hand-tune a value that’s easy to get wrong: too low starves throughput, while too high wastes poll capacity. It’s especially helpful for spiky workloads.
Nexus SDK Support for .NET & TypeScript is now Generally Available. Nexus Operations can now be authored and consumed from the .NET and TypeScript SDKs at GA quality, rounding out cross-language support for calling Temporal-based services via Nexus.

Builder spotlight#

This month, we’re featuring Temporal’s own Andrew Baker, who built imaxxing.io, a website that helps you land IMAX Odyssey tickets before they sell out. The app runs on Temporal and Google Cloud, using Durable Execution to stay reliable through the sudden demand that hits a ticketing tool.

The project is an example of how anyone can run a Temporal Serverless Worker on Cloud Run. Andrew deployed his Workers to Cloud Run and let Temporal Serverless scale them with real usage, reducing resource use when traffic was low and expanding capacity the moment the site went viral. That elasticity matters most during the traffic spikes a ticketing app faces when demand peaks. The best part: the whole thing ran inside the free tier of both Temporal and Google Cloud.

Watch the interview we did with the Cloud Run team to see how Andrew built it.

How to Temporal#

Temporal News in the Web UI#

There’s a new Temporal News feed in the Temporal Web UI. It’s another place to learn about new features, upcoming events, technical articles and videos, as well as exclusive digital swag like the Ziggy coloring sheet. The feed refreshes automatically, so the latest updates are there when you need them.

Look for the megaphone icon in the top right navigation next time you’re using Temporal. Don’t see it? Upgrade to the latest Web UI to access the news feed.

New Standalone Activities tutorials#

Learn how to build a durable job queue with Standalone Activities in new hands-on tutorials for Go, Java, Python, and TypeScript. Explore automatic retries and recovery, idempotency, concurrency controls, Heartbeating, and more through hands-on coding.

Route specialized workloads#

Learn how to direct resource-intensive workloads to the right Task Queues and Worker pools. This Validated Pattern shows you how to separate specialized work based on resource needs, optimize infrastructure, and scale Workers independently.

Recover business processes without restarting#

Learn how to build business processes that recover from permanent failures without starting over. This Validated Pattern shows you how to pause when intervention is needed, resolve the issue, and resume from where the process stopped.

Questions from the field#

A new community member asked about concurrency limits for Activities that call asynchronous APIs. In the follow-up discussion, Temporal’s Cory Finger shared documentation, a custom script example, and plenty of configuration options. If you need a way to build out a solid timeout and Retry Policy for asynchronous work, take a look at Cory’s response.

Matt Muniz da Silva asked whether Task Queue Fairness was generally available in the Java SDK. David Reiss on our engineering team confirmed that it has been since May 2026, but the follow-up discussion uncovered a bug. Community member sragireddy filed a GitHub issue and PR that fixed it. It’s a great example of the community surfacing a real SDK bug, tracing it to its root cause, and shipping a fix within the same thread!

Durable Digest: August highlights