IBM believes over a billion new applications will be built with Gen Ai: Here’s how they will help to happen with an agent AI

Enterprise AI in 2025 passed from experiments to application and implementation, developed by AI assistants to AI agents.

This is the main topic of IBM Think 2025 CoNference that starts today. At the event, IBM Announces an extensive list of new AI services as well as improvements to existing technologies to help relocate AI’s AI efforts to the real -world introduction. The IBM updates core is a series of updates on it Watsonx platform This was first announced on Think 2023. At the Think 2024 event, the big theme was The Introduction of an orchestration And the ability to help Enterprise build their own AI assistants. In 2025, AI assistants were table bets and the conversation in the industry and in every enterprise is how to build, use and take advantage of agency AI.

IBM announces a series of agent AI capabilities, including:

• AI Catalog: Centralized Center for the opening of pre -constructed agents.
• Connecting the agent: A partnership program for third -party developers to integrate their Watsonx Orchestrate agents.
• Domain -specific templates For sales, orders and HR.
• Creator of agents without code For business users without technical expertise.
• Agents for the development of agents For developers.
• Multiangent orchestra with opportunities for cooperation between agent and agent.
• OPC agent (In a private review) provision of telemetry and monitoring.

The main goal of IBM is to help businesses to overcome the difference between experimentation, real -world implementation and business benefits.

“Over the next few years, we expect more than a billion new applications built with the help of generative AI,” IBM CEO Arvind Krishna said at a press and analysts briefing. “AI is one of the unique technologies that can reach at the intersection of productivity, cost savings and scanning revenue.”

ENTERPRISE AI challenge: How to get a real return on investment

Although there is no shortage of over -interest in AI, this is not what it actually has a real difference for the bottom line.

A study sponsored by IBM shows that businesses receive only the return on investment (ROI), which expects approximately 25% of the time. Krishna noted that several factors influence the return on investment. These include access to business data, the hidden nature of various applications and the challenges of hybrid infrastructure.

“Everyone is doubled from investing in AI,” Krishna said. “The only change in the last 12 months is that people stop experiments and focus a lot on where the business value is.”

From AI experimentation to production of enterprises

At the heart of IBM’s messages, it is recognized that organizations are shifted by isolated AI experiments on coordinated implementation strategies that require opportunities for an enterprise department.

“We are trying to overcome the gap from where we are today, which is thousands of experiments in the implementation of corporate class that require the same type of security management and standards that require the critical applications of the mission,” said Ricito Gunnar, General Manager and AI in IBM, in front of Venturebeat.

The evolution of the IBM Watsonx platform reflects the wider maturity of AI technology. The platform was first announced by IBM in 2023, largely as a way to help build and work with AI assistants and vending machines. In 2024, as the AI ​​agent first began to become the main one, IBM began to add agency capabilities and partnered with many suppliers, including Crew AI.

With the new components of IBM Agentic AI, now the direction is to enable cooperation and work streams with many agents. It is about exceeding the ability to build and deploy agents to understand how an enterprise can generate a return on investment from agents.

“We really believe that we are entering the era of real intelligence systems,” Gunnar said. “Because now we integrate AI that can do things for you and it’s a great distinction.”

The technologies and protocols that allow Enterprise Agentic AI

The industry has no lack of attempts to help activate agency AI.

Langchain is a widely used platform for construction and working agents and is also part of a wider effort along with Cisco and Galileo for the open AGNTCY frame for agent AI. As for Communications between Agent and Agent, Google announced Agent In April. Then, of course, there is Model’s context protocol (MCP), which appeared to become a factual standard for connecting agent AI tools to services.

Gunnar explained that IBM uses its own technology for the piece with many agents. She noted that how agents work together is crucial and is a point of distinguishing IBM. She also said that IBM was trying to take an open approach. This means that businesses can build agents with IBM tools, such as Beai, or those from other suppliers, including Crew Ai or Langchain, and all of them will still work with Watsonx Orchestrate.

IBM also allows MCP. According to Gunnar IBM, it supports MCP by facilitating tools with MCP interface to automatically display and be used in Watsonx Orchestrate. Moreover, if there is a MCP interface tool, it will automatically be available for use in Watsonx Orchestrate.

“Our goal is to be open,” she said. “We want to integrate your agents, no matter what frame you have built in.”

Consideration of enterprise problems: security, management and observance

As part of making sure that the AI ​​agent is ready for use of businesses, it is necessary to guarantee confidence and compliance.

This is also a critical part of IBM’s pressure. Gunnar explained that IBM had built railings and control directly in the Watsonx portfolio.

“We are expanding the opportunities we have to manage LLMS in agency technology,” she said. “Just as we have a LLM rating, you need to be able to evaluate what it means to the agent’s answers.”

IBM also expands its traditional indicators for evaluating machine training by agent. Gunnar said IBM tracks over 100 different indicators for large language models, which it now extrapolated and extended to agents.

Influence in the real world

The AI ​​agent has already been impacting in the real world for many organizations.

IBM uses its own agent AI to help improve its own processes. Gunnar noted that using his own HR agent, 94% of simple to complex requests in IBM is actually answered by HR agent. For the public procurement tasks, the use of IBM from your own agent work flows has helped reduce the award time of up to 70%.

Another large group of organizations that already take advantage of the IBM agency is the partners of the company. For example, Ernst & Young uses the IBM agency to build a tax platform for its own customers.

What does this mean to businesses

For businesses that want to guide the road in the deployment of AI, the IBM Aentic AI agency provides a plan to move from experimentation to deployment.

Just building an agent is not enough. If IBM CEO is right, the future will include thousands of agents working on businesses. Organizations will build and consume agents and agent services such as MCP from many different sources.

IT leaders should evaluate the platform based on four critical factors:

1. Possibilities for integration with existing corporate systems.
2. Management mechanisms for compatible and secure behavior of the agent.
3. Balance between agent’s autonomy and predictable results.
4. ROI measurement options for implementing agents.

Next is businesses to think now about how agents will work together, how they will be sure and managed. The IBM agency ecosystem will appeal to its customers of the enterprise and openness to connect other agent AI systems that hope that organizations will not create another silo.