What's new in ServiceNow Otto for Creator
Brazil Patch 2 and Australia Patch 8 (November 2026 store release)
New to the series
The ServiceNow Otto for Creator Articles Hub collects every release post and build guide in one place.
Overview
The November 2026 store release is live, and the Brazil family is now generally available.
Two things ship for Build Agent. It is supported on SPP in Australia, the EU, and Singapore, and there are two new model options: OpenAI GPT 6.1 Sol and Google Gemini 3.8 Flash. Those are only 2 new highlights when it comes to Build Agent for November.
Our narrow scope is deliberate. Broader feature development is paused this cycle while the team works on quality, stability, and readiness for the January upgrade.
So this post spends most of its length somewhere else. Let's talk about custom rules and skills you can load into Build Agent today; we'll do a quick overview and have several ready for download at the end of the section.
Access
Build Agent on SPP in the EU
What's shipping: Build Agent is supported on SPP in the EU, alongside Australia and Singapore.
Why it matters: SPP instances in the EU have had no Build Agent path at all. If your organization runs inside that boundary, the answer to every Build Agent question so far has been no.
What this means for you: The answer is yes now, with a check attached. Availability and exact capability parity depend on your region, your release, and how your instance is configured, so confirm the specifics for your own instance before you plan work around it.
Models
Two new model options
What's shipping: Build Agent supports OpenAI GPT 6.1 Sol and Google Gemini 3.8 Flash.
Why it matters: Model choice is one of the few Build Agent settings where the right answer depends on your own work. The model that handles a long multi-table build well is not always the one you want for a one-line script fix, and the only way to find out is to run both against something you recognize.
What this means for you: Both appear in the model selector in the chat input, switchable per conversation. If your organization restricts which providers are available, the list reflects that, so check with whoever owns your model provider configuration before assuming a model is there.
Resources
Rules and skills to start from
What's shipping: Nothing new in the product here. Custom rules and skills have been in Build Agent since the August release. What is new is a reference set: six rules and eight skills, with every field value written out and ready to paste into the Studio editor.
Why it matters: Rules and skills are the cheapest way to get Build Agent building the way your organization already builds. Most teams who try them write one or two and then stop, because an empty Instructions field with no example next to it is a hard place to start.
What this means for you: Read them, argue with them, change what does not fit how you work, and test what is left on a sub-production instance before anyone turns them on for a team.
One naming note before the set. These are Build Agent rules and skills, authored in ServiceNow Studio under Settings and applied to an instance, an application, or a single user. They are a different thing from the ServiceNow Otto for Creator skills catalog, which is a set of product capabilities shipped by ServiceNow. Same word, two meanings.
How the two differ
Same editor, same fields, different economics.
|
Rules Loaded into every session Always in context, whether the task needs them or not. Build Agent caps you at five active rules in total today, so each one you add is a choice not to add another. Keep them short and absolute. Best for the guardrails you want on every piece of generated code. |
Skills Called by name, when the work needs one Cost nothing until someone invokes one, which buys room to be long and specific. Best for procedures with judgment in them: design reviews, audits, triage, and the decisions you want settled before anything gets built. |
The five-rule cap is blunt, and it is also a useful forcing function. The set below has six rules in it. One of them has to wait, and which one depends on what your team gets wrong most often.
The skills went a different way. The ones that earned a slot tell you what is wrong or what to decide next: a data model review that stops and asks before creating any tables, an access audit that queries live role and group membership and reports without changing anything, a walkthrough of an application you inherited and nobody documented. Two of them hand the heavy exploration to a sub-agent and return only the findings, which keeps the context in your main session clean.
Worth knowing
A skill fires when a request matches keywords in its Description, so the Description does double duty. It documents the skill and it triggers it. Write it with the words people actually type.
What's in the set
Rules. No hardcoded sys_ids or instance-specific values, server-side query hygiene, business rule type discipline, client scripts stay asynchronous, scope and naming conventions, declare the platform choice before you build.
Skills. Data model review, access control design, performance triage, deployment readiness check, author ATF test coverage, integration design, access audit, explain this application.
One of the rules in full, so you can see the shape before you download anything:
Name: Declare the platform choice before you build
Description: When a request could be implemented with more than one
platform capability, state which one you chose and why in one line
before generating anything, so the user can redirect you cheaply.
Instructions for Build Agent:
Prefer configuration over code, and low-code over script: Flow Designer
over a business rule for multi-step or approval processes; UI Policy
over a client script for show/hide/mandatory; Notification engine over
a script that sends mail; Table Cleaner over a scheduled delete script.
Open with one line: "Building this as <artifact> because <reason>."
Then build.
Never generate a legacy Workflow. Use Flow Designer.That one earns a slot because a show-or-hide requirement can be a UI Policy any admin maintains, or a client script somebody owns forever. The declaration line makes the choice visible while redirecting is still free.
Activating the Explain This Application skill through the chat interface.
These are examples, not a standard
These came out of conversations with developers, architects, and admins about where generated code costs them time later. Every one encodes an assumption about how an organization builds, and at least two encode mine: the naming convention in the scope rule, and the choice of which six rules to write. Test each one against work you recognize, on an instance you can break, before you put it in front of a team.
[ Download the Build Agent rules and skills reference in the .ZIP below ]
Getting started
Build Agent runs in ServiceNow Studio. For this release you will want Brazil Patch 2 or Australia Patch 8. ServiceNow SDK 4.13 is the version paired with this Build Agent release.
Starting with the Brazil release, Build Agent - Trial is active by default. There is no Store install and nothing to request. The trial caps you at 100 prompts, and lifting that cap takes either ServiceNow Otto for Creator or AI Workflow Factory.
What's next
The work this cycle is quality and stability, and making sure the January upgrade is uneventful. That is less interesting to write about than a feature list, and it is what most of the feedback this year has been asking for.
Past that, context handling and repeated work are both still open, along with the permission and governance requests we hear from nearly every customer running a delegated development model. These are still directions, not commitments, but those of you have been following along know we've been prioritizing customer feedback.
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.
Published November 2026. Brazil Patch 2 and Australia Patch 8, November 2026 store release.
