While GenAI has been making headlines, it’s not the job killer many feared. Instead, it’s become a powerful tool that enhances productivity without replacing human roles. The real disruptor lurking on the horizon is agentic AI.
Unlike GenAI and many traditional AI systems, which require human oversight and struggle with complex reasoning, agentic AI is poised to autonomously handle multiple levels of decision-making tasks that were once the exclusive domain of seasoned executives and skilled knowledge workers.
Many AI tools continue to operate in isolation from human expertise, struggling to create a true symbiosis between artificial and human intelligence
While GenAI and other AI systems are making strides, many still operate in isolation from human expertise, failing to establish a true partnership between artificial and human intelligence. This integration gap manifests in several ways:
- Lack of context Integration: Many AI systems, particularly those based on LLMs, operate on a query-response basis without maintaining long-term context or incorporating domain-specific knowledge. For example, a financial analyst using AI for market analysis could receive generic responses that miss the company’s challenges or strategy.
- Hallucination risks: Many GenAI systems are prone to hallucination, generating plausible-sounding but false information. Imagine a manufacturer AI system for supply chain confidently reporting, “Supplier Y experienced a 30% production decrease in Q2 2023 due to labor strikes,” prompting the manufacturer to identify alternative suppliers. In reality, supplier Y had a stellar quarter, but the AI’s lack of real-time data and misinterpreting historical data cost the manufacturer millions in unnecessary supply chain restructuring.
- Limited feedback loops: Some AI tools, especially in rapidly changing environments, are significantly limited when it comes to maintaining context and adapting to real-time feedback. Paradoxically, implementing too many feedback loops can lead to data pollution, where AI systems incorporate artificially generated content into their training data, creating a cycle of self-reinforcing misinformation.
- Handling ambiguity: Real-world business scenarios often involve ambiguous or conflicting information requiring nuanced interpretation. For example, an AI system in supply chain management might struggle to balance conflicting priorities such as cost reduction, speed, and sustainability, optimizing for one factor without considering the subtle trade-offs that an experienced manager would include. This is further complicated by two key human factors: (1) people often have different interpretations of what is “right” based on bias, and (2) people can provide unclear or ambiguous input due to a lack of understanding.
- Opaque reasoning: Many AI systems operate as “black boxes,” providing outputs without clear explanations of their reasoning. In critical domains such as healthcare, this can be problematic. A doctor using an AI diagnostic tool may receive a recommendation without understanding the underlying rationale, making it difficult to integrate this insight with their clinical judgment and effectively explain decisions to patients.
- Limited multimodal integration: Human experts often integrate information from various sources and formats (verbal, visual, textual, intuitive) to make decisions. Most AI tools are limited to processing one or two data types, missing out on human cognition’s rich, multimodal nature.
The next iteration of AI, which HFS calls “Purposeful AI,” represents a significant leap in intelligence and task execution
This evolution is not only about enhancing productivity but also enabling AI systems to take on more complex and autonomous roles within enterprises (see Exhibit 1). A key subset of this evolution to “Purposeful AI” is agentic AI, which is specifically designed to operate autonomously while adhering to human-defined boundaries. Unlike traditional AI systems that require ongoing human supervision, agentic AI can make high-level decisions, manage multi-faceted business functions, and execute tasks independently while continuously learning from its environment and interactions.
In contrast to earlier AI systems that often operated in isolation from human expertise, agentic AI can bring enhanced contextual understanding, enabling it to make decisions that align with an organization’s long-term objectives.
Exhibit 1: The evolution of AI in business intelligence and decision-making
Source: HFS Research, 2024
The ongoing evolution of agentic AI in enterprises could cut both operational and software costs
Agentic AI can potentially disrupt the workforce and the enterprise software landscape, creating significant cost-saving opportunities and posing direct threats. For the workforce, roles that humans traditionally handle are now at risk as AI can autonomously analyze data, make high-level decisions, and execute tasks. Additionally, in other cognitive-heavy domains such as strategic planning, AI has the potential to autonomously analyze data, simulate scenarios, optimize operations, and make decisions in real time more effectively than human-driven processes.
On the software front, agentic AI can potentially consolidate multiple functions that typically require a suite of enterprise tools. This integration of capabilities—from financial planning and supply chain optimization to customer service automation—eliminates the need for separate, costly software systems. Enterprises will no longer require multiple licenses or maintenance contracts, threatening the revenue streams of many specialized software providers as companies shift toward a single AI-driven platform. The result is a significant reduction in both software and staffing costs.
Several AI platforms and organizations are emerging to address the evolving landscape of agentic AI
Both large enterprises and startups are pushing the boundaries of AI-driven autonomy and decision-making, transforming how businesses operate.
Established players:
- Cognizant is developing multi-agent AI systems capable of integrating across multiple business functions for improved decision-making by creating a network of autonomous, goal-directed AI entities.
- Salesforce’s Agentforce offers autonomous agents to handle critical sales and service tasks. Built on the Airkit.ai platform, Agentforce enables both out-of-the-box agents and custom agent creation, empowering organizations to automate customer queries and meeting bookings with minimal human intervention.
- IBM is focusing on creating AI systems that handle complex decision-making autonomously, leveraging its Watson platform to develop multi-agent systems for industries such as healthcare, supply chain, and finance.
Startups:
- Mindcorp.ai, an emerging startup, aims to bridge the gap between current and autonomous AI capabilities. Cognition, its AI platform designed for high-value business challenges, uses multi-agent systems with diverse AI “experts” collaborating to solve complex problems.
- Lyzr, another emerging AI player, provides a full-stack framework that creates autonomous AI agents for enterprises. Lyzr agents autonomously manage complex workflows across sales, marketing, and more, with the option for custom agent creation via a no-code platform. At HFS, we are leveraging Lyzr’s capabilities for our “Super Phil” agent—an ask-me-anything AI designed to deliver insights in the style of Phil Fersht, providing timely, strategic analysis.
- rhino.ai is an emerging player in the software development and business process modernization space. rhino-offers generative AI agents that can extract business logic to create and transform applications that can run across any platform.
As this landscape continues to evolve with more major players launching agentic solutions, startups such as Mindcorp.ai, rhino.ai, and Lyzr must differentiate by offering unique capabilities, ensuring cross-platform compatibility, and demonstrating cost-saving benefits to appeal to enterprise clients.
The Bottom Line: While GenAI is changing how we work, it’s agentic AI that poses the real threat to traditional job roles.
Enterprises must adapt quickly to this shift as agentic AI can autonomously handle complex decision-making tasks. This will impact both workforce roles and the enterprise software landscape, reducing the need for repetitive, decision-heavy positions and consolidating software functions under AI-driven platforms.
