Barriers still exist for many enterprises to fully adopt and reap the benefits of agent-based AI.
IBM bets that the blocker does not create AI agents, but manages them in a production environment.
At TechXchange 2025 today, IBM unveiled a number of features designed to fill the gap: Project Bob, an AI-powered IDE that coordinates multiple LLMs to automate application modernization; AgentOps for real-time agent management; and first open source integration Long flow down Watsonx OrchestraIBM platform for deploying and managing AI agents. IBM’s announcements outline a three-pronged strategy to address the interconnected challenges of AI in the enterprise: modernizing legacy code, managing AI agents in production, and bridging the gap between prototype and production.
The company claims that 6,000 internal developers at IBM have used Project Bob, achieving an average 45% raise in productivity and a 22-43% raise in code commits.
Project Bob isn’t just another vibration programmer, it’s an enterprise modernization tool
There’s no shortage of AI-powered coding tools on the market today, including tools like GitHub Copilot and vibration coding tools like Replit, Cursor, Bolt, and Lovable.
“Project Bob takes a fundamentally different approach from tools like GitHub Copilot and Cursor,” Bruno Aziza, IBM vice president of data, artificial intelligence and analytics strategy, told VentureBeat.
Aziza said Project Bob focuses on enterprises and maintains the context of the full repository during editing sessions. Automates sophisticated tasks such as Java 8 to newer Java and framework updates from Struts or JSF to React, Angular or Liberty.
The architecture combines Claude, Anthropic’s Mistral, Meta’s Llama and recently released IBM products Granite 4 models through a data-driven model selection approach. The system routes jobs to whichever LLM is the best fit, balancing accuracy, latency and cost in real time.
“It understands the entire repository, development goals and security standards, enabling developers to design, debug, refactor and modernize code without breaking the flow,” he said.
Of IBM’s early 6,000 users, 95% used Bob to perform tasks rather than generate code. The tool integrates DevSecOps practices, such as vulnerability detection and compliance checking, directly into the IDE.
“Bob goes beyond helping with code – he coordinates analysis across the entire software lifecycle, helping teams deliver secure, modern software faster,” he said.
Project Bob benefits from up-to-date Anthropic partnership
Part of Project Bob is a up-to-date partnership between IBM and Anthropic
The two providers announced a collaboration to directly integrate Claude models into the Watsonx portfolio, starting with Project Bob. The collaboration goes beyond model integration and includes what IBM describes as a first-of-its-kind guide to deploying AI agents in the enterprise.
IBM and Anthropic co-createdA guide to designing secure enterprise AI agents with MCP servers”, focusing on the Agent Lifecycle (ADLC). The ADLC framework provides a structured approach to designing, implementing, and managing AI systems in the enterprise. MCP refers to the Model Context Protocol, Anthropic’s widely adopted open standard for connecting AI assistants to the systems and data they need to work with.
Make it easier to create enterprise-grade AI agents
In addition to Project Bob, IBM announced that it is expanding watsonx Orchestrate with the integration of the open-source Langflow visual agent builder. Langflow is an open source technology led by DataStax, which was acquired by IBM in May this year. The Langflow integration aims to solve what Aziza calls the “production prototype gap.”
“Today, there is no seamless path from open source prototyping to enterprise-grade systems that are reliable, compliant and scalable,” Aziza said. “Watsonx Orchestrate transforms a Langflow-like agent composition into an enterprise-grade orchestration platform adding management, security, scalability, compatibility and operational reliability, making it production-ready for mission-critical applications.”
Aziza explained that Langflow’s integration with watsonx Orchestration provides critical capabilities in addition to the open source tool, including:
Agent life cycle structure: Delivery, versioning, deployment and monitoring with multi-agent coordination and role-based access.
Integrated artificial intelligence management: The built-in watsonx.governance file provides audit trails, explainability of agent decisions, monitoring for bias and deviations, and policy enforcement. Langflow has no native management controls.
Enterprise infrastructure: SaaS or on-premises hosting with data isolation, SSO/LDAP integration and granular permissions. Langflow users must manage their own infrastructure and security.
No-code and pro-code options: Langflow is “low code”. IBM has added a visual, no-code Agent Builder and professional Agent Development Kit to ensure seamless promotion from prototype to production.
Pre-built domain agents: HR, IT and finance agent directory integrated with Workday, SAP and ServiceNow.
Production observability: Built-in dashboards, analytics and SLAs for enterprise support with continuous performance monitoring.
AgentOps and agentic workflows: from build to manage
IBM is also introducing two up-to-date features to watsonx Orchestrate that work with Langflow integration: Agentic Workflows for standardized agent coordination and AgentOps for production management.
Agentic Workflows solves the problem that Aziza calls “brittle scripts.” Currently, developers create agents using custom scripts that break when scaled in enterprise environments. Agentic Workflows provides standardized, reusable flows that consistently sequence multiple agents and tools. This connects directly to Langflow integration. While Langflow provides a visual interface for creating individual agents, Agentic Workflows handles the orchestration layer, coordinating multiple agents and tools into repeatable enterprise processes.
AgentOps then provides management and observability of running workflows. A up-to-date built-in observability layer provides real-time monitoring and policy-based control throughout the agent lifecycle.
The governance gap becomes concrete in corporate scenarios. Without AgentOps, an HR implementing agent could set up benefits and payroll, but teams have no visibility into whether they’re applying policies correctly until problems arise. With AgentOps, every activity is monitored in real time, allowing anomalies to be flagged and corrected immediately.
What does this mean for businesses
Technical debt is something many organizations struggle with and can be a non-trivial barrier for organizations looking to begin agent-based AI implementations. Project Bob’s value proposition is clearest for organizations with significant legacy Java code bases. IBM’s internally measured 45% productivity gain suggests significant acceleration from Java 8 to newer Java versions and framework upgrades from Struts or JSF to newfangled architectures. However, these metrics come from IBM developers working on IBM systems. The critical unknown is whether multi-model orchestration delivers the same results across customer codebases with different architectural patterns, technical debt profiles, and team skill levels.
Langflow integration fills a real gap for teams already using open source agent frameworks. The challenge is not building agents using tools like LangChain, LangGraph or n8n. It adds a layer of governance, lifecycle management, enterprise security controls, and observability required for production deployment.
For enterprises looking to lead the way in AI adoption, IBM’s announcements underscore the fact that management infrastructure is now at stake. With existing tools, you can quickly create agents. Scaling them securely requires lifecycle management, observability, and policy control.
Project Bob is now available in private preview, with wider availability expected in the future. IBM accepts access requests through its developer portal. AgentOps and agent workflow integrations are available now in watsonx Orchestrate, with Langflow integration expected to be generally available later this month.