Enterprises today are extensively experimenting with large language models (LLMs) and generative AI. According to Matt Carbonara, the managing director at Citi Ventures, these enterprises can be divided into two categories. The first category consists of more conservative enterprises that are adopting this technology in a centralized manner. They are establishing centers of excellence and developing policies to guide their experimentation. The second category comprises organizations that face the risk of being threatened if they do not embrace generative AI technology as soon as possible. This is particularly true for the customer service sector, where the impact of this technology is expected to be transformative.
During his participation in a fireside chat at VentureBeat Transform 2023, Carbonara emphasized that the ongoing change in both large enterprises and startups revolves around understanding the implications of this new technology. Enterprises are grappling with questions such as how it affects their operations, what strategies they should adopt, and whether it poses a threat to their existence.
The Significance of Automation in the Era of Gen AI
In the current era of gen AI and increased experimentation, Carbonara highlighted the continued importance of automation. Enterprises are investing substantial amounts of money in automation, which is utilized in various ways. Citi Ventures perceives automation as the utilization of software to automate different processes within a large enterprise. This includes transaction processing, data processing, customer experience, and customer onboarding.
Carbonara divided the evolution of automation into three distinct phases. The first phase, which he referred to as “RPA 1.0,” involved the initial ability of software bots to manipulate digital systems. The second phase, intelligent process automation, added intelligence to these processes. Presently, we are in the phase of “hyper-automation,” which involves performing more complex tasks across multiple systems using multiple technologies. For instance, optical character recognition and natural language processing are employed to analyze and contextualize documents before feeding the data into an algorithm for decision-making.
Challenges Faced by Enterprises in Automation
Carbonara emphasized that data quality is the most significant challenge faced by large enterprises in automation. Acquiring high-quality data and establishing a “golden set of data” are crucial for making informed and strategic decisions. Regardless of whether an advanced LLM or a simple model is used, the output will be compromised if the data quality is poor.
Another bottleneck in the automation process is integrating cutting-edge technologies into legacy systems. Enterprises must assess whether their existing systems can scale and handle the demands imposed on them. Additionally, regulated industries must implement auditability, controls, and governance measures.
The Future of Gen AI Agents in Enterprises
Carbonara predicted that all large enterprises will eventually have gen AI agents performing various tasks. These autonomous agents will interact with one another, such as a software building agent collaborating with a security agent to address a vulnerability. As these agents will have access to data stores, enterprises will need to establish governance frameworks. This includes determining how agents can access data and what actions they can perform with it. Ensuring data quality and governance are in place is crucial to enable the capabilities brought about by these autonomous agents.
The adoption of generative AI and automation in enterprises is driven by the need to stay competitive and transform operations. Enterprises face the challenge of understanding the implications of these technologies and formulating strategies to leverage their potential. Data quality and integration with legacy systems remain key challenges, while the future lies in the deployment of gen AI agents that require robust data governance frameworks.
Leave a Reply