Business value appropriation roadmap for artificial intelligence

Arindra Nath Mishra (Department of Business Management, XLRI, Jamshedpur, India)
Ashis Kumar Pani (Department of Business Management, XLRI, Jamshedpur, India)

VINE Journal of Information and Knowledge Management Systems

ISSN: 2059-5891

Article publication date: 6 April 2020

Issue publication date: 31 May 2021




Artificial intelligence (AI) is deemed to have a significant impact as a value driver for the firms and help them get an operational and competitive advantage. However, there exists a lack of understanding of how to appropriate value from this nascent technology. This paper aims to discuss the approaches toward knowledge and innovation strategies to fill this gap.


The discussion presents a review of the extant strategy and information systems literature to develop a strategy for organizational learning and value appropriation strategy for AI. A roadmap is drawn from ambidexterity and organizational learning theories.


This study builds the link between learning and ambidexterity to propose paths for exploration and exploitation of AI. The study presents an ambidextrous approach toward innovation concerning AI and highlights the importance of developing as well as reusing the resources.

Research limitations/implications

This study integrates over three decades of strategy and information systems literature to answer questions about value creation from AI. The study extends the ambidexterity literature with contemporary.

Practical implications

This study could help practitioners in making sense of AI and making use of AI. The roadmap could be used as a guide for the strategy development process.


This study analyzes a time-tested theoretical framework and integrates it with futuristic technology in a way that could reduce the gap between intent and action. It aims to simplify the organizational learning and competency development for an uncertain, confusing and new technology.



Mishra, A.N. and Pani, A.K. (2021), "Business value appropriation roadmap for artificial intelligence", VINE Journal of Information and Knowledge Management Systems, Vol. 51 No. 3, pp. 353-368.



Emerald Publishing Limited

Copyright © 2020, Emerald Publishing Limited

1. Introduction

Artificial intelligence (AI) is an attempt to understand and build a machine capable of doing intelligent tasks. AI could be pivotal to the second machine age (Brynjolfsson and McAfee, 2016) and help us in mastering our physical and intellectual environment, leading to prosperity for the humankind. We have dabbled in creating a “thinking machine” for several decades. However, recent headway into AI encourages us to seriously consider the projected impact on the world (Kurzweil et al., 1990; Kurzweil, 2006; Agrawal et al., 2017). This burgeoning transformation of our world hints at an immense opportunity for firms. It has been discussed that companies could use AI for optimizing the business departments like operations (Baryannis et al., 2019), marketing (Roos and Kern, 1996; Sterne, 2017) and human resources (HR) (Sivathanu and Pillai, 2018; Stone et al., 2018). AI adoption promises value embedded in its untapped potentials and warrants a detailed investigation (, 2018).

Despite the widespread understanding of the potential of AI, it is not fully understood, which prevent firms from implementing AI-based technology solutions and extract business value from them. Ransbotham et al. (2017) reported that about 85 per cent of the executives believe AI can help them in business. However, only 5 per cent of companies had extensively incorporated AI, and 20 per cent had partially used AI. This points toward a gap between potential and actual adoption. Another survey found that while 83 per cent banks had considered using AI or machine learning in their business, the implementation rate was around 67 per cent (Dougal, 2018). The gap has been explained by Dougal as a lack of knowledge about the application of these technologies into business problems. The knowledge would come through organizational learning, post which value can be extracted. We have differentiated process into three steps – learning, research and development (R&D) and value creation. This exploration is guided by three pertinent research questions (RQs) around deriving business value from AI:


How could organizations develop knowledge in AI?

The approach toward developing technological competencies required to appropriate value from AI would come from organizational learning (OL) (Crossan and Berdrow, 2003; Real et al., 2006). These competencies could be developed through the exploration of new avenues for AI development as well as making exploitative use of AI by incremental innovation:


What are the ways of deriving value from AI for an organization?

An organization can exploit extant knowledge as well as explore new ways of using AI as proposed by March (1991). However, extending the extant discussion on value creation in current technological context could potentially help formulate a robust strategy for delivery value:


What are some themes of current uses of AI in businesses?

Some of the AI-based tools are part of information systems in organizations and could help in improving the processes. If we can highlight how it has impacted departments like marketing and operations, we could show the potential use cases of AI-based technologies in process improvement.

2. Artificial intelligence

We live in a post-industrial, information-based society (Duff, 2004; Webster, 2007). Over the past four decades, information systems have moved from a peripheral to a central one. Transitioning from decision support systems to data warehouse to real-time warehouse to big data analytics, we are eventually moving toward the “cognitive-computing” era (Watson, 2017). AI will drive this new wave of information technology (IT) at the core of information-based systems, thus opened up avenues for the use of AI in analytics, language processing and visual processing. AI could benefit humanity through a massive increase in information processing accuracy and efficiency, unlocking economic and social development (Hall and Pesenti, 2017). Some of the recent applications have been to chatbots (Hill et al., 2015), translation services (Bahdanau et al., 2014; LeCun et al., 2015; Wu et al., 2016; Johnson et al., 2017), robotics and autonomous agents (Dirican, 2015; Reitman, 1984; Tirgul and Naik, 2016) and virtual assistants like OK Google, Siri, Alexa and Cortana (Bushnel, 2018). It is interpolating from these developments that it can be estimated that in the future, machines could replace humans in tasks like driving cars, solving problems or managing logistics (Brynjolfsson and McAfee, 2016). This is achieved by mimicking the way we humans learn. Tasks like reading a newspaper, is essentially recognizing the letters, assembling them into meaningful words and sentences. AI can do the same using algorithms that are broadly called machine learning (ML) algorithms. ML uses statistics for finding patterns in huge amounts of data (Hao, 2018).

AI makes it possible for machines to be able to adapt and solve problems in uncertain domains. However, the way AI may acquire knowledge and makes sense of the world could shape its perception of the world (Sanzogni et al., 2017). This would require careful deliberation of sources and boundaries of knowledge within AI.

3. Classical technology dilemma: exploration vs exploitation

According to March (1991), there are two broad ways to generate value from technology: firstly, through exploitation, which is “extension of existing competencies, technologies, and paradigms,” while exploration is about finding new alternatives. Factors like uncertainty in the results, time of development, novelty of the development compared to the current process could help decide which path to take (Kuittinen et al., 2013). However, over-reliance on one of these strategies could be detrimental as it leads to “competency trap”; hence, it is suggested to strike a balance between both (Liu, 2006). Ambidextrous organizations are those that attempt to balance both (Tushman and O’Reilly, 1996; O’Reilly and Tushman, 2004).

This means that organizations need to have skill sets to survive in mature markets, which play by the rules of efficiency, and new products markets, which play by the rules of innovation, agility and flexibility (Tushman and O’Reilly, 1996). Though there exist competing schools of thoughts who prescribe focusing on one or the other or both, we look into value creation as a function of ambidexterity as it has been shown to work best for survival of new ventures (Hill and Birkinshaw, 2014).

While the competencies and the R&D are integral components of any OL strategy, the major rethinking for AI would be based on the two aspects that have come up in recently: servicitization and open development. Servicitization is turning products into service where possession of physical hardware is with the purveyor who lets the buyers subscribe to them without the hassle and cost of procurement (Ramiller et al., 2008). One of the ways to achieve servicitization is technology development over a platform.

On the other hand, open development, as the name suggests, involves a non-proprietary approach of managing intellectual property. Open innovation has also been termed as external knowledge exploration (Lichtenthaler, 2011). This can come in the form of outsourcing, sharing and collaboration, thereby developing competencies. Open innovation is an emergent method of R&D (Enkel et al., 2009). We have tried to explore the relevance of open innovation in AI R&D.

4. Roadmap for artificial intelligence

We build our roadmap from the empirical findings of Turulja and Bajgorić (2018). They demonstrated that “knowledge” is the precursor to “innovation,” in turn leads to business performance in firms. This broad knowledge innovation strategy can be played out by several combinations, as shown in Figure 1.

There are four different pathways presented, two each for type of knowledge strategy and two types of innovation strategies. As discussed earlier, two approaches to knowledge could be either to apply existing knowledge (exploitation) to solve business problems or seek new knowledge (exploration) to address business problems (March, 1991). The goal of all business processes is to convert inputs into valuable outputs. New technology is expected to improve these processes through innovation. Business value can be created either through new products (product innovation) or through optimizing the business process (process innovation). In the context of AI, the innovation in product comes through new product development (NPD). Another type of innovation would be innovation in process which would be brought about by business process transformation (BPT).

The technology transformation roadmap is indicated as an approach toward business value appropriation from AI as it moves from its current to future state. The current state is the one where the firm has not yet embraced AI for its business, while the future state is the one where it would have embraced and used AI. The three steps of OL, R&D and value creation are not strictly sequential neither chronological. However, Figure 2 depicts a general direction toward the future state, which is dependent on the organizational learning, which would be elemental for the R&D process that would be used for the value creation. This is seen through the dimensions of either exploration or exploitation of technology.

4.1 Organizational learning

OL is defined as “development of insights, knowledge, and associations between past actions, the effectiveness of those actions, and future actions” (Fiol and Lyles, 1985). OL is a continuous, irreversible and path-dependent process (Nieto, 2004), which is a process of encoding learning into a guiding framework for the organization (Levitt and March, 1988). One of the critical aspects of OL is organizational knowledge (OK). OK is the ability to carry out business processes based upon the past collective understanding of the domain and context (Tsoukas and Vladimirou, 2001). It has been shown that OK can serve both as a barrier as well as a source of inspiration for the NPD (Carlile, 2002). On the other hand, there is a strong link between success of BPT and knowledge management practices (Jang et al., 2002).

Exploitation is essentially knowledge application and it gets operationalized in sensing and seizing new opportunities for the firm (Teece, 1998). Exploitation strategy works like an adaptation that can be done quickly and predictable, while exploration has inherent risk, and it is a time-taking process (Real et al., 2006). This could be more suited for firms that needs to implement fast and frugal AI innovation. On the other hand, the exploitation strategy works through knowledge acquisition and represents the firm’s ability to recognize usable knowledge, then endeavor to absorb it (Liao et al., 2009). The development of a proprietary algorithm (one that is more accurate in prediction or faster or computationally efficient than previous algorithms) will lead to a distinctive competency. These types of distinctive competencies (Real et al., 2006) help the organization in developing an inimitable value proposition for the customers, leading to a competitive advantage (Crossan and Berdrow, 2003).

In the case of NPD, primarily, there needs to be a focus on exploration. On the other hand, for BPT, exploitation priority is a better approach. However, in both cases, the complementary approach should also take place to balance out the benefits.

The takeaway for a firm indulging in BPT is that IS competencies lead to increased entrepreneurial agility for the firm (Chakravarty et al., 2013), which in turn leads to higher firm performance (Sambamurthy et al., 2003).

4.2 Research and development

Each company may have different requirement and needs, which would result in differing information systems requirements (Gregory, 1995). R&D serves the core function of exploration, and we will have a look at different kinds of R&D. There are three kinds of R&D setups: decentralized, network and integrated (DeSanctis et al., 2002). Decentralized designs are more suitable for firms that want to improve the existing product. This would be better for firms that are indulging in BPT using AI. It would also be optimum for a firm that is AI producer doing NPD if AI is one of the many products it makes because of the reusability prospects of the research insights.

4.2.1 Open innovation, outsourcing and collaboration.

One of the exploitation methods could be through outsourcing of the process to save time and reduce uncertainty (Kuittinen et al., 2013). This is supported by an increasing trend toward external resources is upheld by the value in sharing the development, deployment and maintenance. Open innovation is an approach suited to such business environments with marked uncertainty in resources like AI. AI applications can be developed using components like codes, open APIs, JSON streams and algorithms from easily available open sources. It has been rightly said that “Most innovations fail. And companies that don’t innovate die” (Chesbrough, 2006). The solution proposed by Chesbrough is open innovation where a firm uses both internal and external ideas that can result in faster product-to-market, lower cost and higher firm sustainability. It was estimated that there were less than 10,000 professionals with the right skill set for AI development (Nott, 2018), which leads to a skyrocketing in the salaries of AI professionals (Metz, 2017). In this landscape, it makes even more sense to have open innovation and indulge in collaborative development.

The second approach toward development is to outsource. Outsourcing is taking external help in filling the gaps in an organization’s IS capabilities. These gaps are a function of the resource attributes and resource allocation (Cheon et al., 1995). AI’s complexity and resource dependencies call for greater need for outsourcing. It has also been shown to reduce effort, increase profitability and reduce risks (Choi et al., 2018; Loh and Venkatraman, 1995). Collaboration can also be a mechanism for the acquisition of external resources. Universities and independent research laboratories can be of help in exploration of knowledge for firms. It has been demonstrated by Paunov et al. (2019) that corporate patent filings had higher citations of university publications that were located near. The best examples are Google, Microsoft and Uber who have collaborated with universities in Canada and the USA to co-develop AI-based technological solutions. Another successful example is the Catapul program in UK that helps link industry and research (Davis, 2015). On the other hand, there are examples like Australia that has low collaboration between industry and the academia (OECD, 2019). The willingness to collaborate could be lower due to the gap between academic and practical perspective or due to lack of direct benefits for the practitioners (Rodríguez et al., 2014).

4.3 Value creation processes

The first decision should be prioritizing on either acquisition or application of knowledge or a mix of both. Second would be the way knowledge is used to create value. This requires R&D activities, which result in innovation. The knowledge processes can lead to two kinds of innovations – product innovation resulting in new product development or process innovation leading to an optimized process. Knowledge has been shown to leads to innovation, which in turn leads to firm performance (Turulja and Bajgorić, 2018).

4.3.1 New product development.

One of the ways of value appropriation is NPD where the firm may indulge in the identification of opportunities where it can enter the market with a new offering. As discussed earlier, exploitation approach deals with the incremental innovation, while explorative approach deals with the disruptive innovation. Incremental innovation approach would improve a business process, while disruptive would be akin to a new process (Norman and Verganti, 2014). However, IT projects are generally complex and have a higher failure rate when compared to other engineering projects (Berti-Equille and Borge-Holthoefer, 2015). It is essential to develop the AI-based technologies in a cooperative manner where there is synergy between IT teams and business units to meet larger goals. It can be further added that skills and knowledge need to be readjusted for adaptation in a flexible manner (Leonard-Barton, 1995).

4.3.2 Business process transformation.

It has been rightly said that “no business survives over the long term without reinventing itself” (Bertolini et al., 2015). BPT can be defined as the transformation of products, processes and organizational aspects. The genesis of BPT may be in the factors that could potentially make the current processes inefficient and thereby threaten the sustainability of the firms. The internet changed some of the business models and slowly brought an end to physical distribution of data.

BPT is essentially a redesign of the business processes with the intention of improvements in cost, quality, service and speed (Hammer and Champy, 2009). The first step toward BPT is the same as that of NPD – “identification of opportunities.” This is a crucial step where the firms may question themselves on certain assumptions and business processes. One of the questions they can ask is, “how could they serve the customer better with new technology?” In the case of AI-based solutions like virtual assistants, the firm may answer the question by saying that it would eliminate the waiting time for customer query, it would standardize the experience and provide faster resolution of queries.

Similarly, we may ask “how would the technology impact our business?” On similar lines, it can be answered as – the use of chatbots would reduce the HR requirements and reduce cost. Another way to question the business model is whether any new technologies could make the current process simpler. The end goals may be improved customer experience, reduced cost of operation, reduced cost of goods sold, improved revenue, enhanced lead-time, higher quality (Jha et al., 2016). This would then serve as inputs in deriving the OL and R&D strategies.

Once the firm decides on what processes to improve, it must consider alternatives if available or benchmark it with existent technology or process to make sure that the technology is worth upgrading. While this evaluative phase is undertaken, the criteria should be very clearly laid out. Based on a review of extant literature, Mitropoulos and Tatum (2000) enumerated five attributes that govern the adoption process:

  1. Compatibility – The “task-technology fit” theory suggests a stronger fit between the technology and the task for higher performance (Goodhue and Thompson, 1995). The application area of AI must be a holistically seen as an interaction with the employees, teams, departments of the organization.

  2. Complexity – It is the level of difficulty associated with understanding the technology like AI that encompasses lots of different types of approaches and entails varying levels of complexities.

  3. Observability – Unless the organization can contemplate upon and scrutinize the alternatives they have at hand, they should not move forward.

  4. Triability – Building upon the previous point, the proof of concept (POC) needs to be tested out. This would reduce the problems later on.

  5. Relative advantage – These POCs are rated and compared based upon their performance, cost and risk as per the importance of these three attributes.

The comparison helps in selecting the best option from the different concepts. Once the choice of technology is finalized, the business transformation process can be undertaken.

5. Contemporary implementation themes

5.1 Servicitization

“Servicization is a business strategy to sell the functionality of a product rather than the product itself” (Örsdemir et al., 2018). The need for establishment and maintenance of hardware and software is eliminated by a centralized service offering which is charged for usage and remains on tap. As discussed earlier, in simple terms, it means converting a product into a service. One of the common servicitization in IT is “cloud computing” where the computing functions like storage and processing takes place on a remote computer available to the user as service rather than as a physical product (Mell and Grance, 2011; Varghese and Buyya, 2018; Belbergui et al., 2017). There is no need to set up, maintain and upgrade computers, thereby helping in cost cutting. It also increases the accessibility and use of resources which can be accessed anytime, anywhere.

Cloud computing is usually classified based upon the service that is provided and named as “X as a service” which is an acronym for “anything as a service.” There are three main types of XaaS, namely, IaaS – Infrastructure as a Service, PaaS – Platform as a Service and SaaS – Software as a Service (Kavis, 2014). IaaS is a type of cloud where the cloud vendor provides only the servers, storage and networking while the client sets up their operating system and software on the vendor’s servers (, 2019). IaaS is most useful for organizations that do not have physical space or infrastructure. PaaS is a cloud platform where server as well as operating system and developmental tools are provided by the vendor. It is useful for rapid and collaborate development of applications. SaaS provides access to applications through network eliminating the need to install software on client devices. This eliminates the hassle on maintaining the hardware as well as software.

SaaS is easiest to implement but least flexible, which makes it more suitable for BPT, while IaaS is more suitable for NPD because of the flexibility and controls it offers at the cost of reduced ease of use. PaaS is somewhere in between both and could be leveraged for both. Apart from these, there are some other Xaas implementations that could benefit AI. Business process management over PaaS (or BPM PaaS) enables execution of customized business processes on the cloud (Riemann, 2015). These preset instructions can help reduce time and effort in managing business processes. BPaaS or “business process as a service” is delivering business process outsourcing (BPO) over cloud (, 2019). Essentially it is automation of human interactive agents like call center respondents.

There are different applications of AI, right from prediction engines to virtual assistants to robotics and each case would fit into an appropriate type of an X as a service platform. A brief overview of three main XaaS architectures: IaaS, PaaS and SaaS and their relevance to AI is provided in Table I where we present some examples of each of the types of XaaS.

An example of IaaS is Amazon’s Elastic Compute Cloud (EC2) service, which is used by Airbnb to analyze over 50 GB of data daily (, 2019). EC2 enables load balancing of huge chunks of static data like user pictures and dynamic data like user activities into different EC2 instances on cloud. One of the examples of PaaS would be Microsoft’s Azure ML Service, which offers algorithms for text analytics (, 2019a). Text analytics algorithms could be used by a retailer to analyze social media data to ascertain whether discussions are positive or negative based on sentiment score (, 2019b). Similarly, a SaaS offered by IBM under Watson Developer Cloud is “Tone Analyzer,” which can be easily integrated with chatbots and social streams to understand the emotions and communication styles (, 2016).

However, we also need to discuss some of the key drawbacks of cloud-based AI technologies. Though cloud would provide lower cost, higher quality of service, scalability and time reduction, it may entail reliability and security issues as well as long-term cost ineffectiveness. There is a lot of discussion around security and privacy of AI and having some sensitive data and processing locally. This is possible by using hybrid cloud (Jain and Hazra, 2019).

5.2 Decentralized value creation

There has been an overall trend toward decentralization of the firm’s role in value creation. The change from being a “content gatekeeper” to “customer gatekeeper” opens up avenues in value creation through content creation, infrastructure, access, modules and orchestration (Pagani, 2013). As per our discussion on the platforms of AI value chain, there can be different kinds of platforms. The second aspect is the shift of the power from the firms to the users. This would mean that consumer have more say in what they want, how they want and when they want. This information may either be explicitly stated by the customer or implicitly stated in how they interact with the products. Consecutively, the firms would need to align to these requirements and develop their strategy from these inputs (Brenner et al., 2014).

6. Artificial inteligence use cases in two managerial areas

6.1 Enhancing the customer segmentation

One of the traditional but powerful approaches in marketing is customer segmentation. Segmentation is segregation of different types of buyers who will respond to different kinds of marketing efforts. Traditionally, it has been successful in enhancing marketing effectiveness; however, AI can help increase marketing accuracy further by delivering one-to-one marketing. Micro-segmentation is classification of customers on a finer level and reveals more nuanced aspects of their preferences, lifestyle and aspirations rather than broader aspects like price-sensitivity. ML algorithms can be used to map the customer journey to understand the patterns like effect of change of location on purchase of luxury products or finding a subset of price-sensitive customers who may purchase specific luxury products. Micro-segmentation can further enhance the personalization of the marketing campaigns (Kushmaro, 2018).

An interesting case for micro-segmentation is that of Boeing Employees Credit Union, which used micro-segmentation to optimize email communications resulting in higher responses to promotion drives for loans, credit cards and mortgages (Rijn, 2019). This resulted in a 10 per cent lift for the campaigns.

6.2 Operations and logistic efficiency

There are many applications of AI in the field of operations (Cohen and Sherkat, 1987; Lau et al., 2009). The use of AI opens up new avenues for managing the operations not just for firms that indulge in the movement of physical goods but also help improve the service quality applying the same principles where it is applicable. Overall, the optimization of operations management (Jacobs et al., 2004) using AI would provide the firm with a competitive advantage over others because of increased performance because of better prediction of the volatility of time because of large number of factors. These factors were difficult to model without using advanced ML techniques.

DHL is one of the leaders in logistic services. They have developed an ML algorithm that uses over 58 parameters to predict the average transit time a week in advance (Gesing et al., 2018). This is one of the many use cases that have shown potential of AI in logistics and supply chain optimization.

7. Discussion and conclusion

AI is an umbrella term for some of the most potent upcoming technologies that could open up new avenues of development of solutions. It has been projected that AI could analyze the data that was previously not analyzable, create real-time insights and enhance a firm’s performance management (Clerck, 2018). Also, it can be applied to automate different kind of processes, leading to reduced labor requirements and enhanced efficiency. However, a key difference between successful use of AI and unsuccessful would lie in the acquisition and development of learning about this technology.

This requires developing technological competencies through OL as well as using the developed competencies to develop products and services. This bifurcation helps in simplifying the focus on development or acquisition of scarce resources. Technologies like chatbots are a great starting point as POC for AI. They are found to be not just a novelty but offer functional benefits (Shawar and Atwell, 2007). It has already been used in many organizations worldwide and gains more human-like abilities in expression and problem-solving, as it learns from the interactions (Hill et al., 2015). In a way, firms can readily exploit it by making incremental changes and implanting it in their enterprise. There are also many enterprises IT solutions making use of AI like the AI-based knowledge mining tools, AI-based pattern recognition and robotic applications. We have also discussed general directions of explorative research in marketing as well as operations where AI has been used. However, as the current applications of AI stand, it is needs breakthroughs to develop “self-awareness” and reflection to go beyond the automation of tasks and pattern recognition. Once we develop a way to model such properties, AI can understand the subjectivity of social milieu to have a collective experiential knowledge (Sanzogni et al., 2017).

Lastly, there are also concerns regarding ethical and moral use of AI, which we have not discussed here. We must not forget that the impetus should be on preserving our core values, our belief systems and sense of well-being for not just humans, but the entire ecosystem in which we thrive. We need to ensure that AI transformation does not happen at the cost of our ability to think, ability to reflect, ability to take stock and ability to keep the wheel of world development rotating. It has been said by Elbert Hubbard: “One machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man.”

8. Implications, limitations and future directions

8.1 Implications for theory and practice

AI has been under development for over six decades. However, the recent developments have poised it for becoming the next big disrupter for the businesses (Deloitte, 2019). Usually, the early adopters gain market share as well as have higher revenues (Dos Santos and Peffers, 1995; O’Connor et al., 1998). We have discussed the approaches toward developing competencies in AI through OL. Though the goals of a theorist and practitioner may be different, a good theoretical model could serve both of them well (Dubin, 1976).

This paper has several implications from a manager’s perspective. Firstly, the roadmap should be useful for the managers who are not sure where to begin. The overview of current AI and OL literature could inform the strategic directions for AI adoption. Secondly, we have shown that knowledge management (KM) should be the starting point while the managers need to consider product or process innovation as mediator between KM and firm performance (Turulja and Bajgorić, 2018). Thirdly, the two broad approaches to innovation through NPD and BPT could help in defining organizational priorities and help in extracting firm performance through innovation. The proposed roadmap is flexible enough to cater to different organizations and could serve as a guiding light in navigating the uncertain landscape of AI adoption and value appropriation.

8.2 Limitations and future directions

Through this paper, we have only touched theoretical foundations of rich literature in strategy and IS. This rudimentary overview of concepts need much deeper understanding and analysis and a deep dive could inform the reader about them in detail. This discussion aimed to provide an executive’s overview of the AI strategy. However, internal technology adoption is challenging. At an operational level, there are many factors that govern the technology adoption and continued usage behaviors (Karahanna et al., 1999), but we have not discussed about the adoption, rather limiting this discussion to learning and R&D. The discussion presented here has been developed from extant literature. However, it could be further developed through an empirical inquiry. Future work can look into this area. Ambidexterity is one of the most difficult tradeoffs for an organization (O’Reilly and Tushman, 2004), and the simplistic presentation of the roadmap should not be taken as a remedy pill and needs due diligence. One of the major challenges in development would be to conceptualize and create AI that can have cognitive awareness, which is best described as “subjective awareness” of the world (Nagel, 1974). Future researchers can provide basis for such holistic approaches that go beyond automation of tasks. There are many challenges in developing AI as well as implementing those developments. This presents many opportunities too for researchers as well as practitioners and they can look into these as future directions.


KM, innovation and performance

Figure 1.

KM, innovation and performance

Technology transformation roadmap

Figure 2.

Technology transformation roadmap

AI Cloud platforms

Cloud model AI solution Service provider
IaaS Infrastructure like storage, processing and AI engine for higher customization Amazon EC2
PaaS Platform to build and deploy AI solutions using reusable codes and APIs Amazon Web Services ML, Microsoft Azure ML
SaaS AI-based apps and software through a subscription model Microsoft Cognitive Services, Watson Developer Cloud (IBM)


Agrawal, A., Gans, J.S., and Goldfarb, A. (2017), What to Expect from Artificial Intelligence, MIT Sloan Management Review. (2019), “Airbnb case study – amazon web services (AWS)”, Amazon Web Services, available at: (accessed 25 May 2019).

Bahdanau, D. Cho, K. and Bengio, Y. (2014), “Neural machine translation by jointly learning to align and translate”, ArXiv Preprint ArXiv:1409.0473.

Baryannis, G., Validi, S., Dani, S. and Antoniou, G. (2019), “Supply chain risk management and artificial intelligence: state of the art and future research directions”, International Journal of Production Research, Vol. 57 No. 7, pp. 2179-2202.

Belbergui, C., Elkamoun, N. and Hilal, R. (2017), “Cloud computing: overview and risk identification based on classification by type”, 2017 3rd International Conference of Cloud Computing Technologies and Applications (CloudTech), presented at the 2017 3rd International Conference of Cloud Computing Technologies and Applications (CloudTech), pp. 1-8.

Berti-Equille, L. and Borge-Holthoefer, J. (2015), “Veracity of data: from truth discovery computation algorithms to models of misinformation dynamics”, Synthesis Lectures on Data Management, Vol. 7 No. 3, pp. 1-155.

Bertolini, M., Duncan, D. and Waldeck, A. (2015), “Knowing when to reinvent”, Harvard Business Review, Vol. 93 No. 12, pp. 90-101.

Brenner, W., Karagiannis, D., Kolbe, L., Krüger, J., Leifer, L., Lamberti, H.-J., Leimeister, J.M., Österle, H., Petrie, C., Plattner, H. and Schwabe, G. (2014), “User, use and utility research: the digital user as new design perspective in business and information systems engineering”, Business and Information Systems Engineering, Vol. 6 No. 1, pp. 55-61.

Brynjolfsson, E. and McAfee, A. (2016), The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies, 1st ed., W. W. Norton and Company, New York, NY, London.

Bushnel, M. (2018), “AI Faceoff: Siri vs Cortana vs Google assistant vs Alexa”, available at: (accessed 7 January 2019).

Carlile, P.R. (2002), “A pragmatic view of knowledge and boundaries: boundary objects in new product development”, Organization Science, Vol. 13 No. 4, pp. 442-455.

Chakravarty, A., Grewal, R. and Sambamurthy, V. (2013), “Information technology competencies, organizational agility, and firm performance: enabling and facilitating roles”, Information Systems Research, Vol. 24 No. 4, pp. 976-997.

Cheon, M.J., Grover, V. and Teng, J.T.C. (1995), “Theoretical perspectives on the outsourcing of information systems”, Journal of Information Technology, p. 11.

Chesbrough, H.W. (2006), Open Innovation: The New Imperative for Creating and Profiting from Technology, Harvard Business Press.

Choi, J.J., Ju, M., Kotabe, M., Trigeorgis, L. and Zhang, X.T. (2018), “Flexibility as firm value driver: evidence from offshore outsourcing”, Global Strategy Journal, Vol. 8 No. 2, pp. 351-376.

Clerck, J.-P.D. (2018), “Corporate performance management – role of AI and new technologies”, I-SCOOP, 16 April, available at: (accessed 12 April 2019).

Cohen, A.I. and Sherkat, V.R. (1987), “Optimization-based methods for operations scheduling”, Proceedings of the IEEE, Vol. 75 No. 12, pp. 1574-1591.

Crossan, M.M. and Berdrow, I. (2003), “Organizational learning and strategic renewal”, Strategic Management Journal, Vol. 24 No. 11, p. 1087.

Davis, G. (2015), “Poor research-industry collaboration: time for blame or economic reality at work?”, The Conversation, available at: (accessed 30 November 2019).

Deloitte (2019), “Artificial intelligence disruption”, Deloitte United States, available at: (accessed 22 April 2019).

DeSanctis, G., Glass, J.T. and Ensing, I.M. (2002), “Organizational designs for R&D”, Academy of Management Perspectives, Vol. 16 No. 3, pp. 55-66.

Dirican, C. (2015), “The impacts of robotics, artificial intelligence on business and economics”, Procedia - Social and Behavioral Sciences, Vol. 195, pp. 564-573.

Dos Santos, B.L. and Peffers, K. (1995), “Rewards to investors in innovative information technology applications: first movers and early followers in ATMs”, Organization Science, Vol. 6 No. 3, pp. 241-259.

Dougal, J. (2018), “Banks excited by AI, but uncertainty remains”, NCR, 19 April, available at: (accessed 23 May 2019).

Dubin, R. (1976), “Theory building in applied area”, Handbook of Industrial and Organizational Psychology, R& McNally College.

Duff, A.S. (2004), “The past, present, and future of information policy: towards a normative theory of the information, society”, Information, Communication and Society, Vol. 7 No. 1, pp. 69-87.

Enkel, E., Gassmann, O. and Chesbrough, H. (2009), “Open R&D and open innovation: exploring the phenomenon”, R&D Management, Vol. 39 No. 4, pp. 311-316.

Fiol, C.M. and Lyles, M.A. (1985), “Organizational learning”, Academy of Management Review, Vol. 10 No. 4, pp. 803-813. (2019), “Business process as a service (bpaas)”, Gartner, available at: (accessed 10 December 2019).

Gesing, B., Peterson, S.J., and Michelsen, D. (2018), Artificial Intelligence in Logistics, Whitepaper, DHL Customer Solutions and Innovation, Troisdorf, p. 45.

Goodhue, D.L. and Thompson, R.L. (1995), “Task-technology fit and individual performance”, MIS Quarterly, Vol. 19 No. 2, pp. 213-236.

Gregory, M.J. (1995), “Technology management: a process approach”, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, Vol. 209 No. 5, pp. 347-356.

Hall, D.W. and Pesenti, J. (2017), “Growing the artificial intelligence industry in the UK”, Independent Review for the Department for Digital, Culture, Media and Sport/Department for Business, Energy and Industrial Strategy, available at: www.Gov.Uk/Government/Publications/Growing-the-Artificial-Intelligence-Industry-in-the-Uk

Hammer, M. and Champy, J. (2009), Reengineering the Corporation: Manifesto for Business Revolution, A, Zondervan.

Hao, K. (2018), “What is machine learning?”, MIT Technology Review, available at: (accessed 30 November 2019).

Hill, S.A. and Birkinshaw, J. (2014), “Ambidexterity and survival in corporate venture units”, Journal of Management, Vol. 40 No. 7, pp. 1899-1931.

Hill, J., Ford, W.R. and Farreras, I.G. (2015), “Real conversations with artificial intelligence: a comparison between human–human online conversations and human–chatbot conversations”, Computers in Human Behavior, Vol. 49, pp. 245-250. (2016), “Watson tone analyzer”, 28 November, available at: (accessed 25 May 2019). (2019), “IaaS PaaS SaaS cloud service models”, IBM, Corporate, 20 November, available at: (accessed 10 December 2019).

Jacobs, F.R., Chase, R.B. and Aquilano, N. (2004), Operations Management for Competitive Advantage, Vol. 64, McGraw Hill, Boston, p. 70.

Jain, T. and Hazra, J. (2019), “Hybrid cloud computing investment strategies”, Production and Operations Management, Vol. 28 No. 5, doi: 10.1111/poms.12991.

Jang, S., Hong, K., Woo Bock, G. and Kim, I. (2002), “Knowledge management and process innovation: the knowledge transformation path in Samsung SDI”, Journal of Knowledge Management, Vol. 6 No. 5, pp. 479-485.

Jha, M., Jha, S. and O’Brien, L. (2016), “Combining big data analytics with business process using reengineering”, 2016 IEEE Tenth International Conference on Research Challenges in Information Science (RCIS), IEEE, pp. 1-6.

Johnson, M., Schuster, M., Le, Q.V., Krikun, M., Wu, Y., Chen, Z., Thorat, N., Viégas, F., Wattenberg, M., Corrado, G. and Hughes, M. (2017), “Google’s multilingual neural machine translation system: enabling zero-shot translation”, Transactions of the Association for Computational Linguistics, Vol. 5, pp. 339-351.

Karahanna, E., Straub, D.W. and Chervany, N.L. (1999), “Information technology adoption across time: a cross-sectional comparison of pre-adoption and post-adoption beliefs”, MIS Quarterly, Vol. 23 No. 2, pp. 183-213.

Kavis, M.J. (2014), Architecting the Cloud: Design Decisions for Cloud Computing Service Models (SaaS, PaaS, and IaaS), John Wiley and Sons.

Kuittinen, H., Puumalainen, K., Jantunen, A., Kyläheiko, K. and Pätäri, S. (2013), “Coping with uncertainty – exploration, exploitation, and collaboration in R&D”, International Journal of Business Innovation and Research, Vol. 7 No. 3, p. 340.

Kurzweil, R. (2006), The Singularity is near: When Humans Transcend Biology, Penguin USA, New York, NY.

Kurzweil, R., Richter, R., Kurzweil, R., and Schneider, M.L. (1990), The Age of Intelligent Machines, Vol. 579, MIT press Cambridge, MA.

Kushmaro, P. (2018), “How AI is reshaping marketing”, CIO, 4 September, available at: (accessed 25 May 2019).

Lau, H.C., Chan, T.M., Tsui, W.T., Ho, G.T. and Choy, K.L. (2009), “An AI approach for optimizing multi-pallet loading operations”, Expert Systems with Applications, Vol. 36 No. 3, pp. 4296-4312.

LeCun, Y., Bengio, Y. and Hinton, G. (2015), “Deep learning”, Nature, Vol. 521 No. 7553, p. 436.

Leonard-Barton, D. (1995), “Wellsprings of knowledge: building and sustaining the sources of innovation”, SSRN Scholarly Paper No. ID 1496178, Social Science Research Network, Rochester, New York, NY, available at: (accessed 11 January 2019).

Levitt, B. and March, J.G. (1988), “Organizational learning”, Annual Review of Sociology, Vol. 14 No. 1, pp. 319-338.

Liao, S.-H., Wu, C.-C., Hu, D.-C. and Tsuei, G.A. (2009), “Knowledge acquisition, absorptive capacity, and innovation capability: an empirical study of Taiwan’s knowledge-intensive industries”, Technology, Vol. 11, p. 13.

Lichtenthaler, U. (2011), “Open innovation: past research, current debates, and future directions”, Academy of Management Perspectives, Vol. 25 No. 1, pp. 75-93.

Liu, W. (2006), “Knowledge exploitation, knowledge exploration, and competency trap”, Knowledge and Process Management, Vol. 13 No. 3, pp. 144-161.

Loh, L. and Venkatraman, N. (1995), “An empirical study of information technology outsourcing: benefits, risks, and performance implications”, p. 13. (2018), “Embrace the uncertainty of AI”, 23 July, available at: (accessed 23 May 2019).

March, J.G. (1991), “Exploration and exploitation in organizational learning”, Organization Science, Vol. 2 No. 1, pp. 71-87.

Mell, P. and Grance, T. (2011), The NIST Definition of Cloud Computing, Computer Security Division, Information Technology Laboratory, National.

Metz, C. (2017), “Tech giants are paying huge salaries for scarce A.I. Talent”, The New York Times, 22 October, available at: (accessed 9 May 2019). (2019a), “Text analytics – azure machine learning studio”, 6 May, available at: (accessed 25 May 2019). (2019b), “Perform sentiment analysis with text analytics REST API – azure cognitive services”, Microsoft Azure Cognitive Services, 21 November, available at: (accessed 12 December 2019).

Mitropoulos, P. and Tatum, C.B. (2000), “Forces driving adoption of new information technologies”, Journal of Construction Engineering and Management, Vol. 126 No. 5, pp. 340-348.

Nagel, T. (1974), “What is it like to be a bat?”, The Philosophical Review, Vol. 83 No. 4, pp. 435-450.

Nieto, M. (2004), “Basic propositions for the study of the technological innovation process in the firm”, European Journal of Innovation Management, Vol. 7 No. 4, pp. 314-324.

Norman, D.A. and Verganti, R. (2014), “Incremental and radical innovation: design research vs technology and meaning change”, Design Issues, Vol. 30 No. 1, pp. 78-96.

Nott, G. (2018), “Don’t worry, world-leading AI expert doesn’t know what AI is either”, Computerworld, available at: (accessed 9 May 2019).

O’Connor, G., Kung, H. and O’Keefe, R.M. (1998), “Early adopters of the web as a retail medium: small company winners and losers”, European Journal of Marketing, Vol. 32 Nos 7/8, pp. 629-643.

O’Reilly, C.A., 3rd. and Tushman, M.L. (2004), “The ambidextrous organization”, Harvard Business Review, Vol. 82 No. 4, p. 74.

OECD (2019), University-Industry Collaboration: New Evidence and Policy Options, OECD Publishing, Paris, p. 120.

Örsdemir, A., Deshpande, V. and Parlaktürk, A.K. (2018), “Is servicization a win-win strategy? Profitability and environmental implications of servicization”, Manufacturing and Service Operations Management, doi: 10.1287/msom.2018.0718.

Pagani, M. (2013), “Digital business strategy and value creation: framing the dynamic cycle of control points”, MIS Quarterly, Vol. 37 No. 2, pp. 617-632.

Paunov, C., Borowiecki, M. and El-Mallakh, N. (2019), “Cross-country evidence on the contributions of research institutions to innovation”, OECD Science, Technology and Industry Policy Papers, doi: 10.1787/d52d6176-en.

Ramiller, N.C., Davidson, E., Wagner, E.L., and Sawyer, S. (2008), “Turning products into services and services into products: contradictory implications of information technology in the service economy”, in Barrett, M., Davidson, E., Middleton, C. and DeGross, J.I. (Eds), Information Technology in the Service Economy: Challenges and Possibilities for the 21st Century, Springer, pp. 343-348.

Ransbotham, S., Kiron, D., Gerbert, P. and Reeves, M. (2017), “Reshaping business with artificial intelligence: closing the gap between ambition and action”, MIT Sloan Management Review, Vol. 59 No. 1.

Real, J.C., Leal, A. and Roldán, J.L. (2006), “Information technology as a determinant of organizational learning and technological distinctive competencies”, Industrial Marketing Management, Vol. 35 No. 4, pp. 505-521.

Reitman, W.R. (1984), Artificial Intelligence Applications for Business: Proceedings of the NYU Symposium, May, 1983, Intellect Books.

Riemann, U. (2015), “Benefits and challenges for BPM in the cloud”, International Journal of Organizational and Collective Intelligence, Vol. 5 No. 1, pp. 32-61.

Rijn, J.V. (2019), “AI finally makes micro segmentation a reality for financial marketers”, The Financial Brand, 22 October, available at: (accessed 21 January 2020).

Rodríguez, P., Kuvaja, P., and Oivo, M. (2014), “Lessons learned on applying design science for bridging the collaboration gap between industry and academia in empirical software engineering”, Proceedings of the 2nd International Workshop on Conducting Empirical Studies in Industry, ACM, pp. 9-14.

Roos, J.G. and Kern, C.F. (1996), Modelling Customer Demand Response to Dynamic Price Signals Using Artificial Intelligence, IET.

Sambamurthy, V., Bharadwaj, A. and Grover, V. (2003), “Shaping agility through digital options: reconceptualizing the role of information technology in contemporary firms”, MIS Quarterly, pp. 237-263.

Sanzogni, L., Guzman, G. and Busch, P. (2017), “Artificial intelligence and knowledge management: questioning the tacit dimension”, Prometheus, Vol. 35 No. 1, pp. 37-56.

Shawar, B.A. and Atwell, E. (2007), “Chatbots: are they really useful?”, Ldv Forum, Vol. 22, pp. 29-49.

Sivathanu, B. and Pillai, R. (2018), “Smart HR 4.0 – how industry 4.0 is disrupting HR”, Human Resource Management International Digest, Vol. 26 No. 4, pp. 7-11.

Sterne, J. (2017), Artificial Intelligence for Marketing: Practical Applications, John Wiley and Sons.

Stone, C.B., Neely, A.R., and Lengnick-Hall, M.L. (2018), “Human resource management in the digital age: big data, HR analytics and artificial intelligence”, Management and Technological Challenges in the Digital Age, CRC Press, pp. 13-42.

Teece, D.J. (1998), “Capturing value from knowledge assets: the new economy, markets for know-how, and intangible assets”, California Management Review, Vol. 40 No. 3, pp. 55-79.

Tirgul, C.S. and Naik, M.R. (2016), “Artificial intelligence and robotics”, International Journal of Advanced Research in Computer Engineering and Technology, Vol. 5 No. 6, pp. 1787-1793.

Tsoukas, H. and Vladimirou, E. (2001), “What is organizational knowledge?”, Journal of Management Studies, Vol. 38 No. 7, pp. 973-993.

Turulja, L. and Bajgorić, N. (2018), “Knowledge acquisition, knowledge application, and innovation towards the ability to adapt to change”, International Journal of Knowledge Management, Vol. 14 No. 2, pp. 1-15.

Tushman, M.L. and O’Reilly, C.A. III, (1996), “Ambidextrous organizations: managing evolutionary and revolutionary change”, California Management Review, Vol. 38 No. 4, pp. 8-29.

Varghese, B. and Buyya, R. (2018), “Next generation cloud computing: new trends and research directions”, Future Generation Computer Systems, Vol. 79, pp. 849-861.

Watson, H.J. (2017), “Preparing for the cognitive generation of decision support”, MIS Quarterly Executive.

Webster, F. (2007), Theories of the Information Society, Taylor and Francis, Hoboken, available at: (accessed 7 May 2019).

Wu, Y. Schuster, M. Chen, Z. Le, Q.V. Norouzi, M. Macherey, W. Krikun, M. Cao, Y. Gao, Q. Macherey, K. and Klingner, J. (2016), “Google’s neural machine translation system: bridging the gap between human and machine translation”, ArXiv Preprint ArXiv:1609.08144.

Corresponding author

Arindra Nath Mishra can be contacted at:

Related articles