Table 2.

Coding our field notes—An example

	Sample Paragraph from Field notes: “It's difficult to show weekly status updates [during the prototype stage in AI projects] to the clients’ team. They [the client, whose BPO work is also taken up by this organization] are interested in business insights from their data [using AI] in a relatively non-technical fashion from day one. To show such non-tech outputs, we need to extract insights from the data and in the beginning it will be difficult because there won’t be any proper logic [for such extraction] without an understanding about clients’ data. In the last few weeks, it’s still ok as we get some understanding about the relevant features and insights to extract from data by then. But they [the client’s team] want this at the beginning itself, also expect us to give a statement of work before we started. That’s why we do some preliminary work to understand how the client’s data is currently processed by the BPO workforce. They evaluate our work based on this statement.” Paraphrased from discussion with a data scientist
	Coding and summarization-in-brief, done by the mentor to guide the ethnographer’s observations during subsequent visits:
		Emic perspective	Etic perspective	Questions/Pointers
Structure	Technological aspects	It is difficult to plan, structure, and phase-out the execution of AI projects during the prototype stage	AI projects call for greater interconnectedness between the business and technology, and their raison-detre is to be able to make smart business decisions	What are the technical and non-technical outputs that the participants talk about? What business insights are the clients interested in?
	Organizational aspects	Translating technical output in a non-technical fashion is critical for ensuring communication flow between vendor and the client	AI project work happens at the cusp of emerging technologies, business process outsourcing, and consultancy work	If billability is related to the work hours and work statements, what kind of problems are posed for billing AI work—since much of it still is unstructured in comparison to SW/BPO work?
Agency	Participants’ viewpoint	The data scientist wants to build models that work optimally for the client’s use case. But for that per se, she needs to have an understanding of the client’s business and data	Data scientists need to be good at both technical, and business aspects of the AI projects	How do the data scientists look at AI projects which augment BPO work relative to their own research projects?