Artificial intelligence (AI) has come to the fore and is now expected to be one of the most disruptive technologies. It is easy to tell that AI will become pervasive in our everyday life. However, people still lack confidence in AI. Since trust is crucial in the development and acceptance of AI, it is essential to design for proper trust in the human-AI relationship to make people benefit from the technological advance. In this project, the challenge of trust in AI is explored in collaboration with Exact. Exact is one of the market-leading business software companies in the Netherlands. Following the trend of automation in the business software industry, Exact comes up with the future vision of Robotic Accounting and integrate AI into their product Exact Online. However, the users seem to stay behind in adopting these automatic features. Through initial interviews, they find that the lack of trust is one critical reason behind the low adoption. Thus, the design assignment is to “find out the reasons that cause a low trust and adoption of users for automation within Exact Online, and design a solution to promote the trust and usage towards Exact existing and to-be designed AI features.” To understand the background and why trust is such a challenging topic in AI, the project starts with literature reviews around AI and trust. Then trust models are studied to build the theoretical structure of what is trust and how is trust formed. Besides the literature review, the context of Exact and bookkeeping is also studied to define the background of this project and guide the design concept. After the analysis of trust theories and context background, two rounds of user research is conducted to know how users trust AI functions in Exact Online, and how will different elements influence their trust towards : the quantitative research and the qualitative research. The quantitative research is based on the theoretical framework of trust to validate and prioritized the elements, while the qualitative research is more explorative and could explain the elements with real experience. Insights are generated from the two rounds of research and guide the solution design. After the research, a strategy pyramid is created for Exact. The strategy pyramid consists of four layers. The first layer is the vision: trust AI like trust your best assistant; The second layer is the strategy: take care of trust for the whole journey; The third layer is the tactical layer: the trust-building guideline. And the last layer is the operationalization: the envisioned product. The last two layers are further designed. The trust-building guideline is designed as a toolkit that could provoke discussion round trust in the AI development process, and the envisioned product is a redesigned version of Exact Online based on the trust-building guideline. Both the two design concepts are being tested, iterated, and evaluated.