top of page
cover v2.png

Project Brief

Beginning Feb 2025, I led the design of ServiceNow’s first the “Vibe Coding” experience Build Agent in ServiceNow IDE, from vision to execution. As the lead designer in a cross-functional agile team of PM, engineering leads, UX researchers, and VX designers, I drove a fast-paced launch under ambiguity that not only redefined the developer experience but also joined and accelerated the company’s broader AI platform transformation.

Part 1: Visionary Demo

Part 2: Design Execution

  • Secured strong executive and stakeholder alignment, validated through a main stage showcase at ServiceNow Knowledge 2025.

  • Resonated deeply with customers, shaped strategic narrative and potentially contributed to multi-billion dollar business opportunities.

  • The Build Agent in ServiceNow IDE launched on Sep, 2025.

  • Pioneered the “Build Agent” design pattern, establishing the first AI-enhanced coding framework now influencing multiple developer tools across ServiceNow.

Visionary Demo

Demo impact2_edited.png

Problem Statement

Building software can be time-consuming. Turning ideas into features we can preview and iterate on is a lot of work.

Take Aways

AI is the future of development. Conversational interaction and agentic collaboration redefine the way we build sophisticated solutions on the ServiceNow Platform.

Design Philosophy

  • Design for simplicity through conversation — treat building apps like a natural dialogue, not a technical burden.

  • Keep humans at the center — confidence comes from guidance, and shared decision-making.

  • Empower with unified intelligence — context-aware AI agents working together unlock reliable and scalable innovation.

Persona.png

Persona

David - A professional ServiceNow developer is handling a more advanced requirement of the application, prefers to work in a professional code-oriented environment.

Design Execution

Delivered the release MVP design within 3 month

screen example.png

Key Design Challenges

Navigating Ambiguity, Building Clarity

The challenge is to navigate through ambiguity when no one has a definitive answer, while transforming the vision into a clear set of core MVP features through thoughtful prioritization. It requires moving forward step by step, gathering input from cross-functional partners, and iterating the design continuously. The goal is to keep the overall direction steady and aligned, while allowing details to evolve and adapt along the way.

Designing Intuitive Experiences for the New Vibe Coding Paradigm

The challenge is to understand how users will interact with a GenAI-powered tool, where vibe coding introduces a completely new experience that many need time to adapt. For designers, it is also a black box — requiring us to explore and learn the potential behaviors and mental models of engineers. Only by uncovering these patterns can we design experiences that feel intuitive, supportive, and truly effective.

Ensuring Scalable and Consistent AI Patterns Across Platforms

The challenge is to ensure the generality and extensibility of this “Build Agent” AI pattern, so that it can be applied across other ServiceNow Creator tools and future use cases. For example, it should scalable to the low-code ServiceNow Studio platform and ATF testing scenarios, while maintaining consistent interactions and a coherent user experience across different contexts.

Designing for LLM Uncertainty: Smoothing and Mitigating AI Disruptions

The challenge lies in addressing the unique uncertainties of LLMs, where outputs can be unpredictable, occasionally buggy, or even enter unintended loops. The goal is to use design to smooth over these disruptions, anticipate potential issues, and in some cases mitigate or compensate for LLM errors, creating a more reliable and seamless overall user experience.

© 2025 by Yifan Deng.

bottom of page