Opportunities and Risks of Agentic AI Organizations

Organizations reaching Stage III will be able to run a large chunk of their processes autonomously. Agentic organizations connected with each other via agents will create Agentic Supply Chains optimizing resources to gain supreme efficiencies. But that doesn’t mean that Agentic AI will automate all processes. Similar to how craftsmanship labor and small specialized furniture producers still exist, companies in the future will automate parts of their processes while retaining highly emotional, complex, and critical tasks for humans. Future competition will hinge on the effectiveness of Intelligent Process Automations (IPA) and the strategic utilization of human intelligence to create a competitive advantage. In this blog we examine the opportunities and challenges these companies will address in their transformational journey.

Opportunities

Speed. Organizations can deliver services generated and operated by Agents with exceptional speed. For instance, in our Accounts Payable (AP) scenario, typical invoice processing may take several days. With the introduction of Agentic AI, this process could be reduced to mere minutes, with the only bottleneck being Human-in-the-Loop (HITL) interventions. This increased speed will provide organizations with a significant planning advantage as well.

Planning. The principal benefit of Enterprise Resource Planning (ERP) systems and other transactional systems lies in their ability to offer an integrated view of business operations, thereby enhancing planning capabilities. However, this advantage is contingent on timely and accurate data entry. By capturing both unstructured and structured data at the moment it is generated—potentially through a chain of Agents across an “Agentic Supply Chain”—organizations can seamlessly process and record this data into Systems of Record (e.g., ERPs). This capability offers a distinct planning advantage over traditional methods. Planning systems that utilize faster inputs can run machine learning-based scenarios, creating competitive edges in dynamic industries such as Logistics or Retail. Combining near real time data with transactional data give to another AI technology -Machine Learning- input for planning optimization.

Cost. The cost reduction potential for companies employing Agentic AI is comparable to the benefits experienced by those transitioning from handicraft production to mass production. Not all companies will benefit immediately or equally. Reducing costs will require organizations to identify (for example with Process Mining ) processes that have high volume or value and make these processes agentic.

Best Practices. Over time, application companies and consultancy firms will develop Agents embedded with "best practices" tailored to specific processes or industries. Organizations will be able to select these Agentic Process Best Practices and quickly adapt them to their unique environments. Eventually, even small businesses will have the ability to "consume" Agents from the cloud, optimizing their organizational performance.

Risks

Standardization: Initially, low-cost labor and tools provided a competitive advantage to countries and organizations that utilized them. However, with a high degree of standardization, it became challenging to distinguish between different organizations and products. Organizations employing standard Agents will have the initial benefits of cost and speed, but their processes will eventually become indistinguishable from others. Therefore, organizations must decide which processes to standardize using off-the-shelf IPAs and which ones to customize to create differentiation.  

AI Failure: As with any new technology, there will be numerous project and technology failures in the early stages. Just as early robot-manufactured products had high failure rates, Intelligent Process Automations (IPAs) will also experience significant failures, particularly in automating service processes. Effective planning, human-in-the-loop (HITL) approaches, monitoring, and focusing on low-risk processes are crucial strategies to mitigate AI failures.

Environmental Impact: AI is not inherently environmentally friendly. Utilizing complex tools to address simple issues is not an eco-friendly practice. While AI may currently appear cost-effective, its deployment in Agentic Organizations necessitates achieving better results and ROI than current organizational setups. It is imperative to consider the environmental footprint alongside the monetary Total Cost of Ownership (TCO).

Data Security: AI Agents deployed to make decisions and actions on critical IT systems increase the risk of cybersecurity. AI Agents should be designed from the beginning with high security in mind. Security AI Agents should monitor and control the behavior of the execution Agents to ensure there is no breach of data or wrong postings imposed by external parties. On the flip side, many times the securities breaches are happening due to passwords leaks from humans… a strong security policy imposed to AI Agents will minimize this risk..

Regulation: There is a potential risk of government regulations on the use of Agentic AI. Some governments might seek to protect jobs and companies. The transition to Agentic AI could result in layoffs that are difficult to offset in the short term, similar to the industrial revolution. Unlike the past, where workers could retrain for new roles, a human being replaced by Agentic AI cannot easily become a Data Scientist. Consequently, governments may impose taxes on jobs replaced by Agentic AI or introduce other regulatory measures to manage the shift.

Reliance on External AI Companies and Tools: In manufacturing, toolmakers often acquired the expertise to optimize production, while manufacturers became operators. A similar risk exists that a few AI companies might dominate the Agentic AI market. Organizations deploying these technologies might heavily rely on these companies, potentially reducing their profit margins. The shift from industrial to service-based economies was facilitated by outsourcing manufacturing to low-cost countries and creating high-value human jobs to replace lost ones. However, with Agentic AI, there is a risk of outsourcing Agentic processes to specialized companies (a new form of Business Process Outsourcing - Artificial Process Outsourcing), as they may excel in running certain IPAs more efficiently than others.

In the long run Agentic AI organizations will win in the market because they are meeting the basic promise of capitalism – resource optimization. But as with all revolutions, societies should balance how much, how fast and how far this optimization will take place as to ensure humanity will rip the benefits of this technology without risking sacrificing its essence as with the Age of Discovery or with the mass production.