Using Build Agent and AI Agent to cut the cost and time of ServiceNow development

SYSLABO is verifying in house how ServiceNow Build Agent and AI Agent can reduce development cost and shorten development time.

At ServiceNow World Forum Tokyo 2026, we shared what we learned while trying the new features ourselves, along with our trial and error toward putting them to use.

• Reducing time and rework across the whole development process

  Developing business applications takes a lot of time beyond writing programs. Investigating existing systems, organizing requirements, aligning understanding with business departments, and testing and fixing are also essential.

  For example, even for the same “deal management,” the information managed, the timing for finalizing figures, and who can view what differ by organization. If development moves forward while assumptions remain unclear, rebuilding may be needed after implementation.

  In using Build Agent and AI Agent, what matters is not only how fast work can be created, but also whether waiting time for checks and rework can be reduced. SYSLABO is exploring the possibility of rethinking the whole flow of organizing requirements, giving them shape, and checking them.

• Turning requirements into applications with Build Agent

  Build Agent is a development feature that supports creating and changing applications on ServiceNow based on instructions in natural language.

  The official documentation describes a flow in which you explain what you want to build, review the plan Build Agent proposes, and then proceed with creation. You can tell it the users’ roles, the data needed, and the state you want to achieve, and make the result concrete while giving feedback on what is generated.

  What we expect is to shorten the time it takes to turn what is being discussed into something everyone involved can check. If screens and behavior can be checked early, missing information and parts that do not fit the work can be discussed concretely.

• Using it to improve existing applications as well

  ServiceNow in operation also receives many requests, such as adding fields, reviewing processes, and improving screens.

  With Build Agent, you can tell it what to change in the target application and proceed with the changes while reviewing what is generated. Depending on the type of application, preparation such as conversion may be needed before it can be used.

  In making changes, it is important to clarify what to change and what behavior must be kept. A single field may also be used by other processes, reports, or connected systems.

  Understanding why the current specification came about and which work it relates to also helps when giving instructions to Build Agent and checking the changes.

• Thinking with AI Agent, from checking information to carrying out processes

  AI Agent in ServiceNow is a mechanism for retrieving and analyzing information and carrying out processes according to a given role and instructions.

  In AI Agent Studio, you can set roles and the tools to use, and test how agents behave. Configurations in which multiple AI agents share the work toward one goal are also available.

  In development work, collecting necessary information, organizing records, and checking against conditions are tasks where using AI Agent can be considered. To actually hand them over, you need to design the information and tools to refer to and the conditions for judging a task complete, and check how the agent behaves.

  The official documentation also emphasizes clarifying the work an agent handles and preparing tools and descriptions that fit the purpose. Deciding how far an AI agent proceeds and where it hands over to a human check is the starting point for building it into business.

• Putting business goals and conditions into a form AI can use

  Who uses it, and in what situations? What do they want to be able to decide? What exceptions must it handle? Thinking about instructions for AI is also a chance to organize these premises.

  Even for systems that handle the same information, the information needed and how it is shown change depending on whether the system supports a person’s daily tasks or helps a manager judge the outlook for the whole department.

  ServiceNow’s official guidelines also recommend organizing requirements before implementation, making a clear plan, designing data and screens, and giving instructions with specific terms.

  An important part of development is for people who understand the business and people who understand how ServiceNow is structured to work together and express goals and conditions in a form that Build Agent and AI Agent can use.

• Checking generated content and turning it into fixes

  Applications generated by Build Agent also need to be verified against the work. We check whether information is saved correctly, whether only the right users can view and update it, and whether missing input and failed processes are handled.

  The official documentation also describes a flow in which, in addition to reviewing changes and differences, Build Agent helps create and run tests with ATF, the Automated Test Framework, in supported environments.

  While this may reduce the burden of repeated checks, people still need to decide what to test and what state counts as correct, in light of the purpose of the work.

  If important conditions are missing from the tests, a successful result alone cannot show the quality. To make good use of AI support, it is important to be clear about what should be checked.

• Checking cost and time across the whole development process

  Even if implementation becomes faster, the overall effect on development is small if fixes and checks take time. When checking the effect, it is important to look at everything from preparation and implementation to review, testing, and fixes.

  If only certain people can handle the work, a burden remains for later changes and handovers. We also want to check whether the effect can be reproduced for similar work, and whether other members can understand it and keep improving it.

  In our in house verification, we are identifying the tasks where Build Agent and AI Agent are easy to use and the situations that need human knowledge and judgment, and thinking about ways to bring them into actual development.

• Bringing what we have learned to our clients’ development and improvement

  SYSLABO uses Build Agent and AI Agent itself, facing situations where results are not what we expected, and thinks about how to give instructions, how to check results, and how to build these tools into existing development and operations.

  Each client has different work, existing environments, departments involved, and operating structures. We believe that understanding this background and identifying where AI can be used leads to results.

  Aiming to reduce the burden of development and checks so that clients can take on improvements that have been waiting to start, we will apply what we have learned in house to supporting our clients’ ServiceNow development, continuous improvement, and in house development.