Add TechGraph on Google
Artificial Intelligence has already had a significant impact on businesses and has made data-driven decision-making an essential strategy to adopt for business executives. The recent advent of Large Language Models (LLMs) and Generative AI has accelerated this trend and made it possible for AI to enable more far-reaching transformations of business roles.
In what follows we will focus mainly on the impact of LLMs and Generative AI on business and business roles and what we should look forward to in the coming days in this field.
The first area we will examine is financial back-office processing. Robotic Process Automation (RPA) is a technology that has been around for more than two decades. It allows organizations to use software robots (“bots”) to automate routine, rule-based tasks that are carried out in interaction with digital systems. These bots can perform tasks like data entry, form filling, file manipulation, and system navigation. However, RPA bots are usually developed by human workers using process analysis and manual coding.
The current trend is to go beyond traditional RPA by using LLMs and multimodal models to observe and record work processes and convert them directly to software-based automation. Thus, the role of human workers is transformed from that of process observers and process recorders to that of assisting and supervising AI agents that transform an existing back-office human-machine workflow to a purely AI-based agentic workflow.
Another set of major changes is taking place in the field of human resource management. Recent advances such as LLM-based resume-sorting and interview evaluation for recruitment, multivariate classification for performance assessment and training needs formulation, and the use of LLM-based evaluation of performance reports for team formation and work resource planning have transformed the traditional HR role from a subjective and experiential manager to a data-driven and AI-assisted decision maker.
Marketing has also been transformed significantly by LLMs. Content generation, in both textual and image formats, can be carried out very well by LLM and multimodal models and many creative teams have already incorporated Generative AI as a human-machine collaboration tool in the content generation process.
A second major use is the management of projects and campaigns using both LLM tools and more traditional Reinforcement Learning approaches. Another important application area is product idea brainstorming and formulation, and product design, using Large Visual Models.
Finally, lead generation and scoring using automated communication tools that incorporate LLMs for media production are being developed actively and may soon form a core section of the sales process.
The business role is likely to be transformed most significantly the customer service and customer relationship management. As we know, question-answering is a core capability of most LLMs, and the larger models are especially adept at it. This basic feature often combines with “Retrieval Augmented Generation” (RAG) to produce a conversational AI system for answering customer queries.
This means that the front-line customer service representatives are fully automated LLM agents who have substantial knowledgebase to use for question answering and the role of human customer service representatives is transformed to that of managers who are called in only to respond to queries that are found to be too complex and beyond the responding ability of automated agents.
The final role that we will describe is an even more profound change than any of the above. Routine back-office tasks, resume sorting, drafts of creative content generation, and front-line customer service are all examples of business roles that are relatively easy to automate and execute using Generative AI. Even more difficult tasks like complex document processing, employee performance assessment, marketing campaign management, and responding to complex customer queries are well-formulated tasks that may not require truly innovative reasoning.
However, the final task that we will describe is a very complex and open-ended procedure that requires a high level of cognitive performance and is usually carried out by senior management. This is the task of strategic planning and management based on historical and market data, a strong understanding of firm capabilities and resources, and the competitive landscape.
The advent of Generative AI may allow partial automation even of this very high-level function. A proposed model of such a strategy planning system is a multimodal AI combined with a knowledge graph of the enterprise and its industry landscape. These knowledge-based would be constructed using a graph-RAG-based model built using multiple documents and resources about the enterprise and the industry.
The system uses a multi-agent discussion format to discover strategic options and their likely outcomes and risks, then carries out a final discussion in order to arrive at a ranking of various options. This same format can also be used to think about sources of systemic and strategic risks and risk mitigation ideas.
Such a system has been proposed in the AI literature, but it is not yet available for commercial applications. Once such a model is developed and tested, the role of C-suite executives, the CEO, and the Board will change substantially since this kind of strategy-AI will become an intrinsic part of boardroom deliberations and the role of top management will change accordingly.