Accenture
AI Design Playbook
Year
2024
CLIENT
ACCENTURE
SERVICES
SERVICE DESIGN, STRATEGY
While at Accenture, I co-founded the Creative Intelligence lab and developed an experimental design process for collaborating with AI. Creative Intelligence is a design process and framework I created for curious designers like myself working in the age of AI and machine learning. It breaks down the traditional design process, making it more modular and experimental.At its core, it is cross-functional, reflecting the need to combine design and data, art and engineering, human and machine—two sides that have historically been siloed but are now coming together to inspire breakthrough ideas.
在埃森哲工作时，我和同事一起创办了 Creative Intelligence Lab，还搭建了一套和 AI 协作的实验性设计流程。我原创的 Creative Intelligence 这套设计框架，专为 AI 时代乐于探索的设计师打造。我把传统设计流程拆解重组，让它变得更模块化、更具实验性。它以跨领域协作为核心，融合设计、数据、艺术、工程、人力与机器。以往这些领域彼此割裂，如今相互结合，不断催生出突破性的创意。
The Great Interface Shift
What we are witnessing is a creative partnership like never before —one where machines can interpret human needs and respond accordingly and where humans can better steer machines to desired outputs
界面的大转型
如今我们正迎来全新的人机创意协作模式：机器能读懂并响应用户需求，我们也可以更顺畅地引导 AI，拿到想要的设计成果。
Challenge
Old methods like traditional design thinking won’t work here. We need to update our design process for the AI paradigm
挑战
传统design thinking已经不够用了。我们针对AI场景，优化了整套设计流程。
Solution
We need to design with a philosophy that embraces experimentation with AI. Combining data and design helps us find creative answers to complex challenges.
解决方案
设计上秉持勇于用 AI 尝试探索的思路。结合 data 与设计，就能为复杂难题想出新颖解法。
Creative Intelligence Framework
I created this modular system to transform our research and design processes with AI. It has 3 steps
Use AI large models to source extensive big data about the problem.
Combine machine learning techniques and design methods to dig deep into the root causes.
Use AI-powered design workflows to iteratively design rapid, exploratory iterations
创意智能框架
我搭建了这套模块化体系，用 AI 升级调研与设计流程，一共分为三步：
运用 AI 大模型收集大量相关 data，整理问题信息
结合 machine learning 技术与设计方法，深挖问题根源
依托 AI 设计流程，快速产出并优化各类探索性方案
Example 1: Intelligent research focus groups
We can build smarter research workflows by using machine learning to synthesise user data and find hidden patterns for faster design insights. AI can also analyse facial expressions, voices and physical reactions to gauge users’ true opinions of products.
案例1：智能调研焦点小组
我们可以搭建更智能的调研流程，借助 machine learning 整合用户数据、挖掘潜在规律，快速梳理设计思路。AI 还能分析面部表情、语音和肢体反应，判断用户对产品的真实想法。
Example 2: AI Customer Persona
We can build interactive customer personas from real interviews, test new concepts with them and get predictive outcomes.
案例2：AI 客户虚拟形象
我们可结合真实访谈与用户画像，打造可交互的客户虚拟形象，以此测试新创意并获得预判结果。
Collaboration partner: Vodafone
My co-founder and I explored various AI-powered research innovation ideas and wanted to test them with a real client. We pitched our concept to British telecom brand Vodafone and secured a collaboration opportunity to solve their employee attrition challenge. Vodafone was facing severe staff turnover, which destabilized internal business operations. Our Creative Intelligence Lab was tasked with identifying the root causes of employee dissatisfaction and delivering targeted design solutions. I conducted relevant market research and built out a complete AI research plan for the project.
客户合作：沃达丰 (Vodafone)
我和联合创始人构思了多项 AI 创新调研方案，希望落地到实际项目中。我们向英国电信品牌沃达丰完成提案，成功达成合作，协助其解决员工流失问题。当时公司人员流动率居高不下，业务运转受到影响。我们创意智能团队负责挖掘员工不满的根源，并制定针对性设计方案。我完成了相关市场调研，同时搭建了整套 AI 调研方案。
AI Personas
We trialled using AI to turn survey data into “AI personas”. Firstly, I designed and ran a survey across 150+ employees, then analysed responses using design thinking and AI-assisted clustering, tagging, and pattern detection to surface key pain points and behavioural patterns. We were able to generate 2 AI personas based on the full dataset — effectively grounding the models in the survey response data. Next, I built an intelligent knowledge layer that enabled conversational querying of the dataset through these personas. These employee personas acted as “living summaries” of the research, allowing us to prompt, ask questions and quickly surface patterns without manually reviewing hundreds of responses.
AI 虚拟用户画像
我们尝试利用 AI 将问卷数据转化为AI personas（AI 虚拟画像）。我面向 150 余名员工设计并开展调研，结合设计思维，搭配AI clustering（AI 辅助聚类）、标签标注、模式识别完成数据分析，梳理出核心痛点与行为特征。基于这套系统化分析成果，我们依托全部调研数据生成 2 组 AI 虚拟画像，模型完全以 150 余份问卷样本为训练基础。我搭建了intelligent knowledge layer（智能知识层），支持通过虚拟画像以对话形式调取、查询整套数据。这类员工虚拟画像相当于调研数据的动态整合载体，无需人工逐一翻阅海量问卷，只需发起提问、交互问询，就能快速挖掘数据规律
Visual Storytelling
I created a circular visual map to showcase employee pain points and business insights. We refined it together with teams across the company through workshops and co-creation. Following a priority matrix structure, the most severe issues are highlighted in red at the centre, while less critical items radiate outwards.
视觉叙事
我制作环形可视化图谱，展示员工痛点与业务洞察。我们联合各部门，通过工作坊与共创形式优化内容。图谱采用优先级矩阵（Priority Matrix） 架构，最突出的问题以红色标注在中心，次要问题由内向外辐射分布。
Vibe coding
Based on feedback from Vodafone, we focused the direction around the theme of progression, asking: how might we encourage career development and growth? I used Qwen AI to vibe-code early ideas, then refined them in Figma.
Vibe coding
结合沃达丰的反馈，我们将方向聚焦于员工职业发展，并提出思考命题：如何助力员工成长与进阶？我借助千问AI进行 vibe-code 产出初步构想，随后在 Figma 中完成设计优化。
Brand Value Proposition
We aligned on a key shift: progression through a jobs portal alone wasn’t enough — we needed to build a sense of community. This reframed the challenge to: how might we enable smoother collaboration across the business? We explored a broader ecosystem at Vodafone, which we called the “Vodaverse”. Working closely with marketing, we developed the brand narrative “Build your own future”.
品牌价值定位
我们明确核心思路转变：仅依靠职位门户无法满足发展需求，还需打造社群氛围。由此重新梳理课题：如何提升企业内部协作效率？我们为沃达丰规划了全新生态体系，并将其命名为 “Vodaverse”。我与市场团队紧密协作，打造出品牌主题 Build your own future。
Result
This is just the start of the Vodaverse. While the project concluded, we laid the foundation for an AI-powered employee portal that can evolve into a wider ecosystem of intelligent productivity tools.
结果
Vodaverse 的探索才刚刚起步。项目收官之际，我们搭建起 AI 员工门户的基础框架，该平台后续可逐步拓展为综合性效率工具生态。
That's a wrap
Thanks for reading