What's New in Now Assist for Creator: Australia GA Release (June 2026 Store Release)
Improved flow automation, update set visibility, and broader file upload support in Build Agent.
Table of contents
- Overview
- The road to here
- Flow automation in Build Agent
- Update set visibility from chat
- Broader file upload support
- Build Agent in ServiceNow Studio
- Closing the loop on testing
- Improved context capabilities
- Model support
- Resources for builders
- Getting started
- What's next
New to the series? The Now Assist for Creator Articles Hub collects every release post and builder guide in one place.
Overview
The June 2026 store release is live, and it builds directly on the Australia GA milestone from May. This one is focused on Build Agent: deeper flow automation, clearer visibility into what lands in your update sets, broader file upload, and a wider footprint across ServiceNow Studio.
Before we get into it, a thank you.
To everyone who connected with our product and engineering teams at Knowledge 2026, we appreciate the time. The feedback was direct and generous, and it validated a lot of where the roadmap is heading. A particular thanks to everyone who stopped by the Vibe Code Lounge to put Build Agent through its paces. Watching people build real things on the floor, and hearing what worked and what did not, is the best input we get.
Knowledge 2026 Vibe Code Lounge participants building a simple 'Gear Request' app using Build Agent.
The road to here
If you are new to Build Agent, here is the short version of how we got here. Build Agent arrived with Zurich as a conversational developer agent in the ServiceNow IDE. Australia Early Access in March, and the April store release expanded it to ServiceNow Studio.
Australia GA in May was the big one: Test Agent for the full test lifecycle, in-app agents, the MCP client, semantic search, and a set of Studio enhancements. If you missed the May GA post, it is worth a read. Several of this month's features pick up right where May left off.
Flow automation in Build Agent
What's shipping: Build Agent can now create and edit Flows directly, including the advanced patterns that real automation relies on, which means more automation, which is the magic of the platform. So, what flow logic can Build Agent create?
- Parallel execution, so independent operations run at the same time, like dispatching several notifications or updating multiple records at once
- Conditional branching with if, elseIf, and else, using encoded-query conditions and data pill comparisons
- forEach loops over record sets, with exitLoop to break on the first match and skipIteration to filter inline
- try/catch error handling, so a failed action falls back to recovery logic instead of breaking the run
- Subflows with typed inputs and outputs, invokable across applications
- Custom actions that compose the out-of-box steps (create, update, lookup, email, script) into reusable units with their own error evaluation and output assignment
- Flow variables and flow stages for accumulating collections across iterations and tracking named phases of progress
Advanced flow logic generated using Build Agent, explained to a novice developer.
Why it matters: Building Flows in ServiceNow Studio is much faster and simpler with Build Agent. It is a quick way to start a new flow, edit an existing one, or get the scaffolding in place before you take it the rest of the way. And it covers the full range of patterns real automation relies on: branching, iteration, error recovery, parallel work, and modular composition.
What this means for customers: You describe the automation in plain language and get composable, reusable pieces back. Subflows you can call from other apps, custom actions you can reuse, and error paths that hold up. It is a fast way to stand up new automation or extend what you already have.
Update set visibility from chat
What's shipping: Now you can see the update sets Build Agent created right in the chat panel. Each checkpoint in the conversation links to the update set it produced and opens it in a new tab.
Why it matters: Missed update set entries are one of the most common causes of broken deployments. With everything Build Agent creates showing up as it lands, you always know what got captured. The closest analogy is git status, but for ServiceNow metadata, and without leaving the panel to check.
What this means for customers: Less guessing about what is in your update set, and fewer surprises at promotion time. You can trace any change back to the moment it was made. We'll have some improvements coming soon to this area of work.
Broader file upload support
What's shipping: Build Agent now accepts more than just images (tada!). Alongside the existing image formats (PNG, JPG, GIF, SVG, WebP), you can upload PDFs and CSVs, and code and text files including JavaScript and TypeScript (.js, .ts, .jsx, .tsx), JSON, XML, HTML, CSS, Markdown, plain text, and ServiceNow Fluent files (.now.ts). Build Agent reads what you attach, whether that is reviewing a script, parsing a data file, interpreting an error screenshot, or pulling requirements out of a document.
Why it matters: Requirements rarely arrive as a clean prompt. They show up as PDFs, spreadsheets, screenshots, and existing code. Dropping those straight into the conversation means the source material becomes the spec, so you spend less time retyping it into instructions.
What this means for customers: Simply hand Build Agent a requirements document, a policy, a sample data file, or the script you already have, and it works from the real artifact. For example, a procurement policy PDF can become an enforced set of business rules. An error screenshot can drive a fix. A schema can shape a table. Work smarter with natural language and documents alongside to personalize or create new applications on the platform.
Build Agent in ServiceNow Studio
Two of this release's improvements bring more of Build Agent into ServiceNow Studio.
MCP server support in ServiceNow Studio
What's shipping: External MCP server connectivity now extends from the ServiceNow IDE into Studio. The K26 launch partners (Figma, Prisma, Linear) were already live.
June adds Atlassian Rovo, Docusign, Miro, Zoom, Zoom Chat, Zoom Docs, Zoom Revenue Accelerator, and Zoom Whiteboard. We are excited to welcome our new partners and will be adding several more in the months to come.
Why it matters: It brings the external tools your teams already use into the build lifecycle, inside Studio, for platform developers and app builders.
What this means for customers: You can connect Build Agent to the systems where your designs, issues, documents, and conversations already live, without leaving Studio.
Turning on MCP server support in ServiceNow Studio.
UI validation in Studio
What's shipping: The UI validation tool is now available in Build Agent within ServiceNow Studio. In May, UI preview came to Studio. June runs the validation pass there too. After each install, Studio loads the real page in a headless browser and shows it inside, and even includes console errors and broken network calls (if present) to help with troubleshooting. To refine the result, open the preview, click the element you want to change, and choose Edit with Build Agent. That scopes the conversation to just that element, so you can shape it in place and watch it render live.
Why it matters: The edit, save, switch to the browser, hard refresh loop is gone. You see the same pixels an end user would, in the place you are already working.
What this means for customers: Faster iteration on the parts that matter, without rebuilding the whole page or fighting a browser cache.
Editing a single element on a UI page directly in ServiceNow Studio.
Closing the loop on testing
What's shipping: At the end of an app build or edit, Build Agent now prompts you to create and run ATF tests. This builds on Test Agent, which shipped at GA in May to author, execute, and troubleshoot tests across the full lifecycle. June adds the prompt at the right moment, so validation happens before deployment rather than after something breaks.
Why it matters: ATF is powerful but tedious, so tests often do not get written. Most teams do not have someone whose job is maintaining them, and the people building apps would rather build than test. Prompting at the end of a build closes that gap while the work is fresh.
What this means for customers: Test coverage becomes part of the build instead of a separate task you mean to get to later. The tests are real ATF tests, saved to the standard tables in your app scope, ready to run on the same schedules as anything written by hand.
Improved context capabilities
We've mentioned it lightly in this post already, but it's worth further expanding that Build Agent does more in this release to anchor on what you, the developer, are looking at, and to pull in the content you need without leaving the conversation. In May, it started prioritizing what you had open on screen. June takes that further.
Active app scope context
What's shipping: When you open Build Agent from inside an application, while viewing a record or table in that app, it binds that application as the active scope for the entire conversation. A context chip in the panel shows which app is active, so Build Agent already knows the app you mean before you say a word.
The difference shows up in the back-and-forth. Without it, a prompt like "update my business rule" earns two questions first: which app, then which rule. With the active scope set, Build Agent knows you mean the app you launched from, queries the business rules scoped to it, and gets to work.
With the scope set, Build Agent:
- Scopes its queries automatically, so searches for tables, script includes, and business rules filter to the right app without you asking
- Places generated Fluent files in the correct directory for that app
- Builds and installs to the right application
- Skips the app picker and the "which app?" step
A couple of honest limits: the scope is per conversation, so a new conversation means setting it again, or just launching from inside the app. You cannot switch apps mid-conversation; to work on a different app, start a new one. And the scope does not limit what Build Agent can read. It can still query other tables for references. It only decides where new metadata gets created and installed.
Why it matters: The old loop was a round of "which app, which record, which field" before any real work began. Anchoring on what you have open removes that, the way a colleague sitting next to you would already know which app you mean.
What this means for customers: Shorter, more natural prompts. "Add a field to the equipment table" instead of "add a field to the equipment table in the Equipment Catalog app." Less time restating where you are, more time describing what you want.
The active app context chip shows which script Build Agent is scoped to.
Search retrieval tool support
What's shipping: Build Agent can now fetch and present indexed content, including knowledge articles and catalog items, directly within the agentic workflow.
Why it matters: It keeps the agent grounded in the content already published on your instance, without sending you to a separate search interface to find it.
What this means for customers: References and answers come back in the flow of the conversation, so you stay in the build instead of context switching to look something up.
Model support
You can now select Anthropic Claude on AWS, Opus 4.6, alongside Gemini Pro 2.5 and OpenAI 5.4, and pick whichever best suits the task. It is toggled in Now Assist Settings, under Manage Model Providers.
Resources for builders
Now that the release is in your hands, two guides will help you get more out of Build Agent.
The Build Agent Best Practices Guide covers how to structure prompts, ground the agent in your instance, and get reliable results on real work.
The Prompting Guide works from a simpler premise: ask Build Agent to teach you.
And if you missed the Australia GA release in May, that post has the full picture on Test Agent, in-app agents, the MCP client, and semantic search.
Getting started
Build Agent runs in the ServiceNow IDE and ServiceNow Studio. To use this release you will want to be on Australia Patch 2 OR Zurich Patch 9 or later. UI validation uses Cloud Runner, so make sure that is available on your instance.
As a reminder, you can try Build Agent without an entitlement. The Build App - Trial store app gives each account 100 prompts per month, and personal developer instances come with Build Agent activated at 25 prompts per instance. Everything described here is on the standard Build Agent entitlement and the Build Agent - Trial path PDIs are subject to a separate release cadence.
What's next
A few directions we are working on for upcoming releases: continued improvements to update and working sets, more help from Build Agent on implementations, further enhancements to ServiceNow Studio, and improvements to custom rules and markdown handling. These are directional rather than commitments, but they are where our attention is going. We are also combing through all the incredible feedback from Knowledge 2026 and incorporating the guidance on prioritization you provided.
If you build something with this release, tell us about it. Drop a comment below, and let us know what is working and what you want next. That feedback is what shapes the roadmap.
Visit the Now Assist for Creator Articles Hub
Published June 2026. Australia GA, June 2026 store release.
