TLDR: A new focus is emerging in the AI landscape on developing ‘agentic AI’ capable of executing complex, multi-step workflows and making contextual decisions, moving beyond the content generation capabilities of traditional generative AI. Industry leaders emphasize the need for sophisticated infrastructure, human-in-the-loop strategies, and continuous skill development to successfully integrate these autonomous agents into enterprise operations. Gartner predicts a significant rise in agentic AI adoption, with 33% of enterprise software applications expected to include it by 2028.
The evolution of artificial intelligence is entering a new phase, shifting from generative AI (genAI) that excels at content creation to ‘agentic AI’ – intelligent agents designed to execute complex, multi-step tasks and make decisions within context. This progression is highlighted by industry experts who underscore the significant work and strategic planning required to enable these AI agents to truly ‘get stuff done’ in real-world enterprise environments.
Adam Famularo, CEO of WorkFusion, explains that while genAI is proficient at generating content, it often ‘can’t take action’ and struggles with ‘hallucinations and lack of real-time performance.’ Agentic AI, in contrast, is built to act with intent, learn from outcomes, and leverage a foundation of machine learning, statistical analysis, robotic process automation, and intelligent document processing. These agents are envisioned as ‘automations that can execute complex, multistep workflows, across digital environments and apply reasoning.’
However, the deployment of these advanced AI systems is not without its challenges. Famularo cautions that an agentic AI architecture requires ‘proper controls and guardrails.’ Nitesh Bansal, CEO of R Systems, adds that ‘the shift to agentic AI requires a more sophisticated infrastructure and mindset,’ noting that ‘many organizations lack AI literacy, specialized skills, and market ready tools and infrastructure.’
A report from Harvard Business Review Analytic Services, underwritten by Wipro, reinforces these points, stating that agentic AI’s ‘implications stretch far beyond automation or productivity gains.’ The report also emphasizes that ‘preparing an organization and its people to adopt agentic AI can be challenging as leadership may face disinterest, skepticism, or resistance.’ A key takeaway is the enduring importance of ‘critical thinking’ as a human skill, regardless of AI’s advancements, as agentic AI is expected to augment, rather than replace, human capabilities.
Market projections indicate a rapid acceleration in agentic AI adoption. Gartner estimates that by 2028, a substantial 33% of all enterprise software applications will incorporate agentic AI, a significant leap from less than 1% in 2024.
To successfully navigate this transformative period, businesses are advised to:
Get comfortable with AI: Famularo encourages widespread use of tools like ChatGPT and Gemini to understand their capabilities, recognizing they represent only a fraction of broader AI offerings.
Keep employee skills refreshed and current: Bansal stresses the importance of training programs focused on AI technologies and customer service applications to bridge knowledge gaps and foster innovation.
Build use cases: Organizations should identify opportunities for both large and small AI implementations to achieve early value and success. A ‘pre-defined purpose’ approach for AI agents can help avoid overly long development cycles.
Keep humans in the loop: Famularo highlights the necessity of adapting to ‘unknown unknowns’ and ensuring that if an AI agent is ‘confused on the decision to make,’ it can provide data back to humans for resolution.
Treat AI agents as teammates: The HBR report suggests a new era of ‘human teams and AI agents operate in tandem,’ requiring fresh approaches to processes, management, governance, and workforce planning. Famularo envisions ‘fusion teams of AI agents and humans,’ citing Moderna’s merger of tech and HR as an example.
Open the agentic AI development process to a broad base: A collaborative approach can mitigate bias and enhance the utility of AI’s analyses and decision-making.
Rethink the processes AI agents are touching: Famularo advises that processes must support ‘autonomy, collaboration, and real-time adaptability’ as agents make decisions and learn from outcomes.
Also Read:
- Enterprise Data Infrastructure Lags as Autonomous AI Agents Emerge
- AI Agents Revolutionize Finance, Purchasing, and Procurement Operations in 2025
An exciting future development is the concept of an ‘internet of agents,’ where AI agents collaborate across organizational boundaries. Famularo points to the financial sector’s interest in ‘secure, cross-organizational intelligence sharing’ to identify fraud patterns and sanction risks using anonymized data, indicating a growing trust in these evolving agentic systems.
