Agentic AI transforms enterprise workflows as governance challenges emerge for 2026

AI agents arrived in 2025 - here's what happened and the challenges ahead in 2026

In artificial intelligence, 2025 marked a decisive shift. Systems once confined to research labs and prototypes began to appear as everyday tools. At the center of this transition was the rise of AI agents - AI systems that can use other software tools and act on their own. While researchers have studied AI for more than 60 years, and the term "agent" has long been part of the field's vocabulary, 2025 was the year the concept became concrete for developers and consumers alike. AI agents moved from theory to infrastructure, reshaping how people interact with large language models, the systems that power chatbots like ChatGPT. In 2025, the definition of AI agent shifted from the academic framing of systems that perceive, reason and act to AI company Anthropic's description of large language models that are capable of using software tools and taking autonomous action. While large language models have long excelled at text-based responses, the recent change is their expanding capacity to act, using tools, calling APIs, coordinating with other systems and completing tasks independently. This shift did not happen overnight. A key inflection point came in late 2024, when Anthropic released the Model Context Protocol. The protocol allowed developers to connect large language models to external tools in a standardized way, effectively giving models the ability to act beyond generating text. With that, the stage was set for 2025 to become the year of AI agents. The milestones that defined 2025 The momentum accelerated quickly. In January, the release of Chinese model DeepSeek-R1 as an open-weight model disrupted assumptions about who could build high-performing large language models, briefly rattling markets and intensifying global competition. An open-weight model is an AI model whose training, reflected in values called weights, is publicly available. Throughout 2025, major U.S. labs such as OpenAI, Anthropic, Google and xAI released larger, high-performance models, while Chinese tech companies including Alibaba, Tencent, and DeepSeek expanded the open-model ecosystem to the point where the Chinese models have been downloaded more than American models. Another turning point came in April, when Google introduced its Agent2Agent protocol. While Anthropic's Model Context Protocol focused on how agents use tools, Agent2Agent addressed how agents communicate with each other. Crucially, the two protocols were designed to work together. Later in the year, both Anthropic and Google donated their protocols to the open-source software nonprofit Linux Foundation, cementing them as open standards rather than proprietary experiments. These developments quickly found their way into consumer products. By mid-2025, "agentic browsers" began to appear. Tools such as Perplexity's Comet, Browser Company's Dia, OpenAI's GPT Atlas, Copilot in Microsoft's Edge, ASI X Inc.'s Fellou, MainFunc.ai's Genspark, Opera's Opera Neon and others reframed the browser as an active participant rather than a passive interface. For example, rather than helping you search for vacation details, it plays a part in booking the vacation. At the same time, workflow builders like n8n and Google's Antigravity lowered the technical barrier for creating custom agent systems beyond what has already happened with coding agents like Cursor and GitHub Copilot. New power, new risks As agents became more capable, their risks became harder to ignore. In November, Anthropic disclosed how its Claude Code agent had been misused to automate parts of a cyberattack. The incident illustrated a broader concern: By automating repetitive, technical work, AI agents can also lower the barrier for malicious activity. This tension defined much of 2025. AI agents expanded what individuals and organizations could do, but they also amplified existing vulnerabilities. Systems that were once isolated text generators became interconnected, tool-using actors operating with little human oversight. What to watch for in 2026 Looking ahead, several open questions are likely to shape the next phase of AI agents. One is benchmarks. Traditional benchmarks, which are like a structured exam with a series of questions and standardized scoring, work well for single models, but agents are composite systems made up of models, tools, memory and decision logic. Researchers increasingly want to evaluate not just outcomes, but processes. This would be like asking students to show their work, not just provide an answer. Progress here will be critical for improving reliability and trust, and ensuring that an AI agent will perform the task at hand. One method is establishing clear definitions around AI agents and AI workflows. Organizations will need to map out exactly where AI will integrate into workflows or introduce new ones. Another development to watch is governance. In late 2025, the Linux Foundation announced the creation of the Agentic AI Foundation, signaling an effort to establish shared standards and best practices. If successful, it could play a role like the World Wide Web Consortium in shaping an open, interoperable agent ecosystem. There is also a growing debate over model size. While large, general-purpose models dominate headlines, smaller and more specialized models are often better suited to specific tasks. As agents become configurable consumer and business tools, whether through browsers or workflow management software, the power to choose the right model increasingly shifts to users rather than labs or corporations. The challenges ahead Despite the optimism, significant socio-technical challenges remain. Expanding data center infrastructure strains energy grids and affects local communities. In workplaces, agents raise concerns about automation, job displacement and surveillance. From a security perspective, connecting models to tools and stacking agents together multiplies risks that are already unresolved in standalone large language models. Specifically, AI practitioners are addressing the dangers of indirect prompt injections, where prompts are hidden in open web spaces that are readable by AI agents and result in harmful or unintended actions. Regulation is another unresolved issue. Compared with Europe and China, the United States has relatively limited oversight of algorithmic systems. As AI agents become embedded across digital life, questions about access, accountability and limits remain largely unanswered. Meeting these challenges will require more than technical breakthroughs. It demands rigorous engineering practices, careful design and clear documentation of how systems work and fail. Only by treating AI agents as socio-technical systems rather than mere software components, I believe, can we build an AI ecosystem that is both innovative and safe.

From assistance to autonomy: How agentic AI is redefining enterprises

Artificial intelligence (AI) has long promised to change the way enterprises operate. For years, the focus was on assistants, systems that could surface information, summarize documents, or streamline repetitive tasks. While valuable, these technological assistants were reactive: they waited for human prompts and provided limited support within narrow boundaries. Today, a new chapter is unfolding. Agentic AI, whose systems are capable of autonomous decision-making and multi-step orchestration, represents a significant evolution. These systems don't just assist, they act. They evaluate context, weigh outcomes and autonomously initiate actions, orchestrating complex workflows across functions. They adapt dynamically and collaborate with other agents in ways that are beginning to reshape enterprise operations at large. For leaders, this shift carries both opportunity and responsibility. The potential is immense, but so are the governance, trust and design challenges that come with giving AI systems greater autonomy. Enterprises must be able to monitor and override any actions taken by the agentic AI systems. Shift from assistance to autonomy Traditional AI assistants primarily respond to queries and perform isolated tasks. They are helpful but constrained. Agentic AI pushes further: multiple agents can collaborate, exchange context and manage workflows end-to-end. Imagine a procurement workflow. An assistant can pull vendor data or draft a purchase order. An agentic system, however, can review demand forecasts, evaluate vendor risk, check compliance policies, negotiate terms and finalize transactions. It does this all while coordinating across global business departments, including finance, operations and compliance. This shift from narrow support to autonomous orchestration is the defining leap of the next era of enterprise AI. It is not about replacing humans but about embedding intelligence into the very fabric of organizational workflows. Rethink enterprise workflows The goal of every enterprise department is focused on efficiency, scale and standardization. But agentic AI challenges enterprises to think differently. Instead of designing workflows step by step and inserting automation, organizations now need to completely reimagine and architect intelligent ecosystems for orchestrating processes, adapting to evolving business needs, and enabling seamless collaboration between humans and agents. That requires new thinking. Which decisions should remain human-led, and which can be delegated? How do you ensure agents access the correct data without overstepping boundaries? What happens when agents from finance, HR and supply chain must coordinate autonomously? The design of workflows is no longer about linear handoffs; it is about orchestrated ecosystems. Enterprises that get this right can achieve speed and agility that traditional automation cannot match. Accelerate agentic AI-led transformation with a unified platform In this environment, unified platforms become critical. Without them, enterprises risk a proliferation of disconnected agents working at cross-purposes. A unified approach provides the guardrails with shared knowledge graphs, consistent policy frameworks and a single orchestration layer that ensures interoperability across business functions. This platform-based approach not only reduces complexity but also enables scale. Enterprises don't want dozens of fragmented AI projects that stall in the pilot stages. They want enterprise-grade systems where agents can collaborate securely and consistently across the enterprise. Unified platforms simplify outcome monitoring and strengthen governance -- both critical as systems become increasingly autonomous. Build trust and accountability As AI systems act with greater independence, the stakes rise. An agent who makes flawed decisions in customer service may frustrate a client. An agent that mishandles a compliance process could expose the enterprise to regulatory risk. That's why trust and accountability must be designed into agentic AI from the start. Governance is not an afterthought; it is a foundation. Leaders need clear policies defining the scope of agentic autonomy, transparent logging of decisions, evaluating and monitoring agents and escalation mechanisms when human oversight is required. Equally important is cultural trust. Employees must believe these systems are partners, not threats. This calls for change management, training, and communication that positions agentic AI as augmenting human capability rather than replacing it. Measure business value early One of the most common pitfalls in enterprise AI adoption is the gap between promising pilots and at-scale results. Studies show that a significant percentage of AI projects never make it past experimentation. Agentic AI cannot afford to fall into this trap. Enterprises must measure business value early and continuously. This includes efficiency gains, cost reductions, error avoidance and even intangible benefits like faster decision-making or improved compliance. Success will be defined by automation coverage across processes, reductions in manual intervention and the ability to deliver new services at speed and scale. When designed responsibly, agentic AI can deliver exponential improvements. A procurement cycle reduced from weeks to hours, or a compliance review automated at scale, can fundamentally alter enterprise performance. Preparing for the future The rise of agentic AI does not mean handing over control to machines or codes. Instead, it marks the next phase of enterprise transformation, where humans and agents operate side by side in orchestrated systems. Leaders should start by piloting agentic systems in well-defined domains with clear governance models. From there, scaling across the enterprise requires investment in unified platforms, robust policy frameworks, and a culture that embraces intelligent automation as a partner in value creation. The enterprises that succeed will be those that approach agentic AI not as another tool, but as a strategic shift. Just as ERP and cloud once redefined operations, agentic AI is poised to do the same, reshaping workflows, governance, and the very way decisions are made. Agentic AI is moving the enterprise conversation from assistance to autonomy. That change comes with objective complexity, but also with extraordinary promise. The foundation for success lies in unified platforms that enable enterprises to orchestrate with intelligence, govern with trust, and scale with confidence. The journey is just beginning. And for enterprise leaders, now is the time to lead with vision, responsibility, and ambition. N Shashidhar is VP and Global Platform Head of EdgeVerve AI Next. Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they're always clearly marked. For more information, contact [email protected].

The dual reality of agentic AI's inflection point in 2025

When a US-based healthcare provider started using AI agents to assist patients during onboarding and prioritize cases based on severity of symptoms, it reduced pressure on hospital staff and allowed them to focus on more urgent cases. Similarly, several private banks in India are using AI agents for processing loans, to automate underwriting, and document verification. In the telecom sector, Reliance Jio is using autonomous AI agents for real time network maintenance and customer experience workflows. The telco has seen a significant improvement in network uptime, reduction in workload on call centers because of it. These are some of recent use cases that illustrate the rapid proliferation of AI agents across verticals globally as well as in India. According to a November report by EY, 24% of enterprises in India are already using agentic AI. A SolarWinds report claims that over 50% organizations in India plan to adopt agentic AI by 2026. While GenAI has led to a significant boost in individual productivity, agentic AI promises an even bigger growth opportunity for enterprises. By automating complex, multi-step workflows with minimum human oversight, agentic systems can accelerate decision making and reduce turnaround time. Unlike generative AI that relies on prompts from humans to perform a task, agentic AI breaks down complex tasks and executes them autonomously. They are trained to make decisions like humans and solve real time problems using available information and can adapt to new inputs. An AI agent represents a single AI bot that has been trained to perform specific tasks, while agentic AI refers to a framework that includes multiple AI agents. It acts as the orchestrator that breaks down a complex request into smaller tasks and then assigns them to AI agents with specific skill sets. "Enterprises are no longer satisfied with AI that waits for instructions. They are demanding systems that understand context, anticipate intent, and act with confidence inside real operational environments. That shift became unmistakable in 2025 and fundamentally changed the conversation around what enterprises will expect from AI in 2026," said Ganesh Gopalan, Co-Founder and CEO of Gnani.ai. According to Gartner, 33% of AI use cases will involve agentic AI by 2028. How agentic AI became mainstream in 2025 Many enterprises feel that adoption of agentic AI is no longer optional but a strategic necessity. According to a Cloudera report on enterprise AI agents in 2025, 83% firms want to invest in AI agents to maintain a competitive edge, while 96% plan to expand their use in the next 12 months. McKinsey's State of AI in 2025 report also shows that 62% firms are experimenting with AI agents, but the transition from pilots to production-grade systems has been limited. The surge in adoption was driven by the maturity of agentic AI platforms, availability of advanced reasoning models, and a shift in enterprise goal from supporting workers for productivity gains to full-scale automation. For instance, Salesforce's Agentforce platform, which has eased deployment of agentic AI for several organizations, added several new capabilities like the Command Centre that allows enterprise to manage, track and scale agentic AI systems. It also offers reasoning capabilities, allowing agentic systems to solve problems autonomously and adapt their decision making. In 2025, reasoning models became the new industry standard. OpenAI introduced new reasoning models, including the o1 series, which excels at multi-step reasoning. Anthropic released Claude Opus 4a, a reasoning model with long term memory for complex multi-step tasks and extended thinking. Google's latest Gemini 3 models also showed significant improvements in multimodal reasoning. Integrating reasoning capabilities with agentic AI was key focus areas for AI companies. In November, Google announced an agentic AI driven software development platform called Antigravity. It is designed to help developers focus more on managing agents that can write, test and self-correct codes across editor, terminal and browser. Microsoft and GitHub also added multi-model and agentic capabilities to their Copilot tools. Meanwhile, Salesforce and Google are working on cross-platform AI agents using the Agent2Agent (A2A) protocol for enterprises. Salesforce already claims its agentic AI system resolves 83% of queries per week, reducing escalation to humans by half. Krupesh Bhat, Founder and CEO of Melento (formerly SignDesk), points out that AI agents are moving into workflows because complexity has become unmanageable for humans. "Agents now reconcile data, check policy rules, and prepare actions before humans approve them. The key development was multi-agent collaboration across departments. Risk, legal, and operations could finally work on the same workflow without handoffs breaking," added Bhat. Why governance is key to the success of agentic AI As AI agents become deeply embedded in enterprise workflows, organizations are realizing that its success requires a fundamental restructuring and not just a technology upgrade. According to a 2025 Cloudera survey, 50% of IT leaders in India are struggling to scale deployment of AI agents, primarily due to the perceived complexity and usability challenges. This can be attributed to the fact that many enterprises in India still operate on a mix of cloud apps and legacy systems. For an AI agent to work with a decade old ERP system, firms need to build middleware for every task. Similarly, setting up least privilege access for an AI agent to ensure they can do their job without compromising sensitive data can be a daunting task for IT teams. "AI agents are moving beyond task automation to actively supporting planning, analysis, and execution across functions, which requires enterprises to rethink how decisions are made and how accountability is defined," said Sachin Panicker, Chief AI Officer at Fulcrum Digital. Panicker adds that successful adoption depends on strong data foundations, clearly articulated business objectives, and governance models that balance autonomy with human oversight. "As these agents begin to operate across multiple systems and departments, enterprises that invest in orchestration, change management, and outcome-driven metrics will be best positioned to unlock sustainable value while managing risk and complexity," said Panicker. Like GenAI, AI agents too can hallucinate and generate plausible but incorrect outputs. AI agents can also be manipulated through prompt injection attacks to share sensitive company data. AI agents are also stochastic (inherent randomness) and are prone to handling the same task in a different way, which can affect consistency. To mitigate these risks, experts recommend a governance model where human oversight remains a critical checkpoint in the agentic workflow. According to Bhat, accountability drift is a key concern with AI agents. "When agents talk to each other, ownership can blur. In regulated industries, speed without shared accountability is a liability. Collaborative intelligence solves this by making every agent action visible, explainable, and traceable to a human owner," added Bhat. According to the 2025 Cloudera survey, 53% organizations are implementing processes that include human reviews, diversified training data, and formal fairness audits, while 28% have introduced bias-check measures.

Agents, Robots, and Humans: Building the Future of Work Together

By DebDeep Sengupta The future of work is undergoing a seismic shift, driven by agentic AI. Businesses are now facing the challenge and opportunity of redesigning jobs and workflows to maximize the strengths of both human employees and AI-driven agents. According to LinkedIn data, job skill sets have evolved by 25% since 2015 and are expected to shift by 65% by 2030 globally. Similar trends are emerging across regions including India, where employees and executives see GenAI as a tool for enhancing efficiency, not a job threat. However, this transformation demands a strategic reallocation of responsibilities, prompting organizations to rethink workforce structure and operational models. The changing landscape of work Starting 2025 and beyond, companies must fundamentally restructure how they operate. This means reshaping roles, training employees, and distributing tasks between humans and AI agents. This transition will be led by the C-suite leaders, backed by consultants and operations experts specializing in AI-driven business models and large-scale change management. For enterprises to successfully integrate AI, they must carefully orchestrate how work is divided between humans and machines. Without robust infrastructure, governance, and processes, AI adoption risks becoming fragmented and unsustainable. Likewise, a workplace without clear human roles and structured systems can lead to inefficiency and confusion. The key to a seamless transition to agentic automation lies in building a collaborative AI ecosystem, one where AI agents, RPA robots, and employees work together efficiently while maintaining control, visibility, and accountability. Seamless AI-human collaboration through Orchestration In an enterprise setting, orchestration functions as the central nervous system for automation, ensuring the right AI agents receive the right data, at the right time, to make informed decisions and take meaningful actions. It ensures smooth coordination across business processes, minimizes bottlenecks, and enables automation to scale effectively. By integrating AI agents, human employees, and robots into a unified workflow, orchestration creates a seamless ecosystem where AI handles complex decision-making, robots execute tasks, and humans provide critical oversight. The growing role of AI agents in key industries Agentic AI is poised to reshape industries by automating complex, time-sensitive, and data-driven tasks. Some key areas where this transition is already underway include: Healthcare: AI agents enhance patient care by monitoring information flow and supporting employees in making probabilistic, non-rules-based decisions. Banking: AI agents streamline core banking operations by automating customer data reconciliation, accelerating account opening, and ensuring compliance through real-time monitoring. They analyze transactions for fraud detection, generate regulatory reports, and handle high-volume back-office tasks, improving accuracy, reducing turnaround time, and enabling faster, error-free service delivery. IT and Technology: AI agents streamline development and IT operations by auto-generating code, triaging bugs, and managing CI/CD pipelines. Internally, agents monitor system performance, synthesize user feedback, and assist in product decisions, driving faster innovation and cross-team productivity. Manufacturing: AI agents optimize shop floor operations by monitoring machine performance, predicting maintenance needs, and minimizing downtime through real-time alerts. In supply chain and procurement, agents analyze demand patterns, automate inventory management, and trigger replenishment workflows -- improving efficiency, reducing waste, and ensuring production continuity. Real Estate: AI agents automate property operations by validating invoices, updating SAP systems, and handling tenant communications. They enhance customer experience with personalized updates, predictive query handling, and app-based engagement, resulting in faster response time and more reliable service delivery. These applications are just the beginning. Any industry that relies on complex, labor-intensive and data-driven processes stands to benefit from agentic automation. Leading the transition: What enterprises should do now? Identify the key workforce of the future: Start by identifying roles that will change with automation. Process mining and task mining technologies play a key role in analyzing workflow, identifying inefficiencies, and determining tasks suitable for AI. Go with the best-suited technology: Selecting the right technologies depends on a strong partnership between agents, robots, and humans. In an autonomous business, structural tasks like data entry and transfer are managed by robots, while agents focus on more complex tasks like forecasting, comparison, innovation, and problem solving. Keeping humans in the loop is crucial to ensuring accountability, applying judgment, and aligning automation with business goals. Upskill and reskill your workforce: Businesses need to invest in targeted training programs to help employees manage AI tools and interpret data. Equally important are human capabilities such as creativity, critical thinking, and problem solving. Above all, employees must understand how these changes will boost their productivity and contribute to business goals. Optimize workflow for human-machine collaboration: Businesses need to redefine their processes to maximize both AI and human strengths. This requires synergy across departments, seamless integration, and clearly defined goals. The path forward The role of AI in the workplace is rapidly expanding, and AI agents will play a pivotal role in shaping the future of work. These agents will not only assist employees but also take on decision-making, predictive analysis, and real-time execution of complex tasks. To stay competitive, business leaders must act now. Companies that successfully redesign jobs, upskill employees, and create a well-structured AI-human collaboration -- where agents, robots, and humans work in harmony, will define the next decade of innovation. (The author is Area Vice President for South Asia, and the views expressed in this article are his own)