For developers with preferred IDE's and AI coding tools, ServiceNow's April Build Anywhere announcement offers maximum flexibility. Developers can build on ServiceNow using any front door. The ServiceNow SDK and agent skills now work across Claude Code, Cowork, OpenAI Codex, Cursor, Windsurf, and Antigravity. Work in the tools you know and love, then export to ServiceNow Studio like any other scoped app, with governance, security roles, and data models applied automatically.
Before getting into the details, it’s worth seeing the full flow end to end.
Watch: from prompt to production
As part of the SDK, ServiceNow has open-sourced specific build skills that enable coding agents like Claude Code to create native apps like a ServiceNow developer. It knows what ServiceNow platform objects to build, and how to build them using ServiceNow best practices. It understands the platform, the data model, the logic, and the scoped app structure that ServiceNow expects. Your coding agent writes in Fluent code directly into a sandbox on the ServiceNow platform, meaning the output is platform native by design. At no point do you need to open a ServiceNow instance in a browser. Everything is happening in the tools you know and love.
Where context comes into play
Previously, an app that was first deployed had no awareness of what the platform already knew. (Org structure, data relationships, policies, integrations, etc.). Coming in Q2, the Context Engine will let apps inherit platform context such as org structure, policies, integrations once those integrations are configured. Your AI agents will start grounded instead of from scratch. Think of it like an organizational intelligence layer. The integrations already in place (Armis for asset graph, Pyramid Analytics for BI, data.world for data catalog) will give your app’s AI agents live enterprise context to work with from the very beginning. As you can see in this image, for any given request, the Context Engine takes into account things like policies, roles, and request history to make better decisions.
Free Build Agent calls just went up!
- Customer instances now get 100 free Build Agent calls, up from 25
- PDIs now get 25 free calls, up from 10
- Request a free PDI at developer.servicenow.com
For anyone that has hit limits in the past, this offers more runway to work through your reasoning chains. One thing worth noting: descriptive prompts tend to get better results. The more context you give the agent about what you're building and why, such as the data model, the scope, or the business rule logic, the less back-and-forth you'll need and the more space you’ll have to experiment. Your platform expertise is still the variable that most determines the outcome.
One more thing (worth knowing for internal conversations)
AI, Workflow Data Fabric, and EmployeeWorks are now included by default in every ServiceNow level: Foundation, Advanced, and Prime. AI isn't only for the premium tier anymore. This is part of ServiceNow’s strategy to put the power of AI into the hands of every ServiceNow developer.
What you need
You only need two things before you start:
- Zurich or later (or a PDI, free at developer.servicenow.com)
- The Build Agent SDK installed. See what’s available and how to set it up: Start here
The SDK and agent skills produce the same scoped app structure whether you're working out of your IDE or ServiceNow Studio. ATF tests and a readiness scan are generated either way. See what Build Agent can do.
What to check once it deploys
When an app deployed from an external tool hits your instance, it appears in ServiceNow Studio under your app list like any other scoped app. One thing to check before you do anything else- if your app uses AI agents, open AI Control Tower and confirm that they’re showing up. That your confirmation that observability is wired correctly.
A few things to know:
- Branch management from external tools is currently limited. If you need to manage branches, use the Source Control menu or work directly in GitHub. Don’t try to drive that from your IDE yet.
- If deployment doesn’t complete cleanly, check that your credential record in ServiceNow matches the token you’re using in your external tool. Mismatched credentials are a common reason for a failed deployment.
- Full test deployment pipelines from external IDEs are still evolving. The SDK supports app creation and update flows well today. Automated pipeline management is coming, so keep an eye on the release notes.
Your turn
What are you building in your AI tools with ServiceNow?
We want to see what you're building in Cursor, Claude Code, Windsurf, OpenAI Codex, or whatever tool you think in. The front door doesn't matter as much as what you're building behind it. A prototype, a proof of concept, a workflow you finally got off the backlog, all of it counts. Drop it in the comments.
If you haven't started yet, tell us what you're trying to solve. What's the friction you wish AI could help with?
This stuff moves faster when we share what's working.
