IBM has introduced a range of new enterprise-level artificial intelligence products, focusing on production readiness and smooth integration to accelerate AI adoption in organizations. The announcements, made at IBM’s TechXchange 2025 conference, detail a suite of innovations designed to help companies transition from AI experimentation to secure and operationalized AI. These new tools are built to address the complexities of hybrid cloud environments and improve developer productivity, governance, and workflow automation.
At the heart of the new offerings is a focus on agentic AI, a type of artificial intelligence that can proactively and independently perform complex tasks. IBM’s strategy aims to provide businesses with the tools to build, deploy, and manage these AI agents across various business functions, from software development to customer service. The new products underscore IBM’s commitment to providing a flexible and open AI ecosystem, allowing enterprises to use a combination of IBM’s own models and those from partners like Anthropic, Mistral, and Llama. This approach is intended to help companies overcome the low return on investment that has plagued many early AI projects.
New Agentic AI Framework
IBM’s new agentic AI framework, watsonx Orchestrate, is a central part of the company’s enterprise AI strategy. This platform now includes over 500 tools and customizable, domain-specific AI agents from both IBM and its partners, designed to simplify complex workflows. The framework is built to be tool-agnostic and environment-agnostic, allowing for consistent deployment and governance across different technology stacks. Watsonx Orchestrate allows users to create, deploy, and manage AI agents that can collaborate across business functions.
To ensure transparency and control over these AI agents, IBM has introduced AgentOps, an agentic observability and governance layer for watsonx Orchestrate. AgentOps provides real-time monitoring and policy-based controls, giving businesses a complete view of their agentic systems’ lifecycles. This feature is designed to address the “black box” problem in AI, where it can be difficult to understand how an AI model arrives at a particular decision. By providing this level of transparency, IBM aims to help businesses build trust in their AI systems and ensure they are operating safely and effectively.
AI-First Integrated Development Environment
IBM also unveiled Project Bob, a new AI-first Integrated Development Environment (IDE) designed to speed up and modernize the software development lifecycle. Unlike current coding assistants that primarily focus on code completion, Project Bob is positioned as a task-generating, agentic collaborator that can help developers write, test, upgrade, and secure source code. The IDE allows developers to chat directly with “Bob” inside their text editor, speeding up workflows with AI assistance that understands the enterprise’s architecture.
Project Bob integrates a variety of leading large language models, including Anthropic’s Claude, Mistral, Llama, and IBM’s own Granite models. This multi-model approach allows the IDE to select the best capabilities for a given task while maintaining continuity across sessions. IBM has already seen significant productivity gains from Project Bob internally, with over 6,000 of its own developers using the tool and reporting an average of 45% increased productivity. The project is currently in a tech preview phase with some IBM clients.
Strategic Partnership with Anthropic
A key element of IBM’s new AI strategy is a strategic partnership with Anthropic, which will see the latter’s Claude family of LLMs integrated into IBM’s software portfolio. This collaboration is focused on bringing a higher level of safety, reliability, and security to enterprise AI development. The partnership aims to address the concerns that have so far limited AI’s return on investment for many businesses, with one IBM executive noting that less than 5% of enterprises are seeing a significant return on their AI investments.
Mike Krieger, Chief Product Officer at Anthropic, stated that “Enterprises are looking for AI they can actually trust with their code, their data and their day-to-day operations.” The integration of Claude into IBM products, starting with Project Bob, is designed to provide this level of trust. By embedding Anthropic’s models within IBM’s watsonx governance layer, enterprises can deploy AI assistants with the assurance that data confidentiality and auditability will be maintained. Every AI interaction can be tracked, versioned, and explained, which is particularly important for businesses in regulated industries such as banking, healthcare, and government.
Extending Agentic AI to Mainframes
IBM is also bringing its new agentic AI capabilities to its mainframe systems with the upcoming watsonx Assistant for Z. These IBM Z agents will enable a shift from reactive troubleshooting to proactive, intent-driven system management. The new assistants are designed to coordinate across different agents, tools, and other assistants, understand conversational context, and automate operational processes. All of this is done while maintaining the high levels of security and policy compliance that are hallmarks of the IBM Z environment.
The integration of AI into the mainframe is a significant step, as these systems are the backbone of many of the world’s largest organizations, processing massive volumes of transactions and data. By bringing AI to the mainframe, IBM aims to help businesses unlock new insights and efficiencies from their most critical systems. The new watsonx Assistant for Z, paired with the latest IBM z17 platform, is intended to streamline operational workflows and increase operator productivity for these core systems.
