Agentic AI is no longer a futuristic ambition—it’s happening now and rapidly becoming an imperative for business leaders who want to accelerate innovation, efficiency, and growth to outpace competitors. In fact, IDC research shows over 50% of the enterprise application market is already AI assistant or AI advisor-enhanced, with about 20% incorporating complete AI agents.*
Business leaders familiar with GenAI's ability to answer prompts are now shifting focus to agentic AI—systems that autonomously perform tasks within defined boundaries, driving efficiency and cutting costs. Yet implementation remains tricky. One survey reveals that 46% of proof-of-concept projects never reach production. The culprit? Enterprise AI demands integration with notoriously difficult components: disparate data sources and entrenched business processes. Building truly useful tools like enterprise-grade virtual assistants requires orchestrating this complex mix of data, tools, models, and governance frameworks—no small feat for organizations still finding their footing in the AI landscape.
Tanzu Platform addresses this complexity by streamlining the app development and deployment process with built-in access and control to critical data services, and dev tools, as well as governance, and model brokering services. An AI-ready Platform-as-a-Service optimized for private cloud, Tanzu Platform gives enterprises the means to harness the power of GenAI and agentic AI. With Tanzu Platform, enterprises can extend informed, grounded natural language responses, to their existing business critical applications or build new apps and agents rapidly by securely integrating proprietary data and enterprise systems. This is possible with its advanced AI capabilities such as planning, tool use, memory and action chaining to turn ideas into production-ready code in minutes.
The result is an easy entry point for building more secure, scalable GenAI and agentic AI solutions that align with enterprise needs and compliance requirements. My Tanzu colleagues Michael Coté and Adib Saikali and I discussed the business potential of GenAI and agentic AI during a recent event, Racing Toward AI App Delivery with Tanzu: Navigating the Fast Lane with Intelligence. You can watch here.
The promise of agentic AI
Let’s start with defining agentic AI. It refers to a type of artificial intelligence designed to not only respond to queries but to also autonomously perform tasks and take actions based on user input or environmental conditions. While traditional generative AI operates on a "read-only" model—responding with answers, suggestions, or predictions—agentic AI elevates this interaction by executing tasks autonomously.
Agentic AI is made possible by introducing an additional layer of intelligence to generative AI. When a request is made to an agentic AI, the system follows a multi-step process based on the knowledge or tools the agent has access to. First, the system analyzes the request and breaks it down into smaller, actionable tasks; then it autonomously carries out these tasks, which may involve invoking APIs, calling microservices or executing code; and finally, the AI evaluates the results of its actions, often through human in the middle interaction, and makes adjustments as needed to achieve the desired outcome.
For example, imagine an agentic AI flow for an insurance claim: when a claim is submitted, an agent will orchestrate a series of AI systems working in concert—one module uses computer vision to analyze damage photos while another applies NLP to extract details from forms and accident reports.
What's innovative about this example is the agent's ability to autonomously decide which enterprise systems to query for policy information, how to interpret complex coverage rules, and when to apply specific business logic. The system activates a decision engine that weighs policy parameters against claim evidence, utilizing a fraud detection AI that compares the case against thousands of historical patterns. For straightforward claims falling within defined parameters, the agent independently makes approval decisions, calculates appropriate payouts using predictive models, and triggers the payment process—all while allowing human adjusters to intervene, review, or take control at any stage. This human-in-the-loop capability ensures oversight while still capturing efficiency gains from the AI's ability to handle routine cases from intake to resolution.
To achieve this level of sophisticated functionality, agentic AI requires more than just the basic generative AI toolset. Agentic AI apps need a framework to manage context and memory across interactions, as well as a system for managing API calls and integrating various services, which Tanzu Platform provides.
For enhanced accessibility and scalability of agentic development for enterprise customers, Anthropic, a leading AI and research company, published the Model Context Protocol in the fall of 2024. This created a standard for how AI models can interact with external tools and data resources. Soon after, our Tanzu Spring team created the MCP Java SDK, a Java implementation of the Model Context Protocol. We're proud that this work is now the official Java SDK for Model Context Protocol.
As a result, Java developers can create or leverage existing interoperable third-party MCP servers that agentic systems can then call, eliminating what would otherwise be complex, custom, hard-coded connections across systems adding months to development cycles.
Start early and meet developers where they are
Many of the executives I speak with worry about the lack of AI programming skills in their organization. We've worked to address that very real concern by meeting enterprise developers where they are instead of requiring them to re-skill.
Over the past year, we’ve brought the latest AI technology to enterprise Java developers through Spring AI, which is an integral part of Tanzu Platform. Considering many Java developers use Spring, Spring AI is critical in unlocking business value-–developers can add AI capabilities to their apps without having to learn new languages or tools. You could say that we’re transforming Java developers into agentic developers. We deliver all of the features of a proven and established PaaS, helping developers move quickly from idea to code to production, with the tools to help security, compliance, and scalability already built in. That means faster iteration, less rework, and quicker ROI.
Developing AI applications is a highly iterative process, so those who begin early gain key insights, add more advanced AI features to applications and stay ahead with continuous innovation.
But innovation requires a solid foundation. Without a platform handling day-two AI operations—like security, controls, and scalability—developers end up bogged down by operational tasks. Tanzu Platform can help offload that burden so developers can stay focused on transforming ideas into code and getting code to production.
Tanzu Platform helps businesses keep pace with AI evolution
Agentic AI is rapidly evolving and will transform businesses and customer experience. With Tanzu Platform, we offer customers a way to get started quickly so they can iterate, learn and evolve their AI application strategy. The key is fast delivery and learning by doing. This not only uncovers valuable insights but also helps reshape organizational processes to integrate AI more deeply and effectively.
Many of our customers already have access to this functionality either through Spring AI or the GenAI Tile, depending on their current entitlements. We are proud to be there at every step of our customers’ AI journey and help business leaders realize business value in their early investments. Visit us at Tanzu AI Solutions to learn more. We are the proud sponsor host of Cloud Foundry Day, which will be held at Broadcom’s Palo Alto, CA campus-–come early and attend our free AI Workshop, Platform Engineering Skills for GenAI and Agentic Training.
*IDC blog: The Agentic Evolution of Enterprise Applications, April 4, 2025