Beyond code: The new role of the developer

If you build with AI, this is the week to pay attention. At its developer conference, TechXchange 2025, IBM rolled out a wave of updates for developers—from its new Project Bob coding partner to a major alliance with Anthropic that brings Claude into enterprise workflows. IBM says it will deliver on the promise of turning generative AI from a sandbox experiment into an everyday engineering tool.

Building AI agents has never been easier—the real challenge is deploying them into production and making generative AI actually usable in the enterprise. “Just because something is available doesn’t mean it’s usable,” said Dinesh Nirmal, Senior Vice President for IBM Software, during the opening session at TechXchange. 

With more than 100 years of experience building and deploying technologies for businesses, IBM is uniquely positioned to understand what it takes to drive adoption and build a strong ecosystem.

That’s where IBM’s new partnership with Anthropic comes in, which will bring Claude, one of the world’s most powerful LLMs, into its enterprise workflows.  

“We are in the middle of a transformation where the future of work is being reimagined with agents,” said Ruchir Puri, Chief Scientist at IBM Research. “Unlike traditional applications, agents can start doing real work, orchestrating complex tasks and learning from context. But they must be governed with the same rigor as any mission-critical system.”

For Anthropic, the collaboration is an opportunity to solidify Claude’s role as a trusted enterprise tool, not just a consumer chatbot. 

“Enterprises are looking for AI they can actually trust with their code, their data and their day-to-day operations,” said Mike Krieger, Chief Product Officer at Anthropic.

The collaboration begins with IBM’s new AI-first integrated development environment (IDE), designed to help engineers modernize code, automate testing and maintain compliance: Project Bob. “Bob,” as it is internally known, is a developer’s partner across the entire software development lifecycle, automating tasks throughout the whole process.

It’s not another assistant, IBM said at the conference: it’s built to fit into developers’ workflow, from adapting design to deployment. With Project Bob, developers can automate tasks, modernize legacy code or build something new entirely.  

Project Bob brings together agentic workflows, built-in security and enterprise-grade deployment flexibility. Available only in preview, the tool was launched only four months ago inside IBM but is now used by more than 6,000 developers internally, according to Neel Sundaresan, General Manager for Automation and AI, IBM Software.

Sundaresan said the initial results are promising, with 90% time savings observed on tasks like test generation, boilerplate and dependency updates. 

“When we interviewed our developers, they said they were able to do things they didn’t think they could do before,” said Sundaresan during TechXchange. “When we bring in AI the right way, we can automate the mundane and augment things not possible previously.”

“It’s not about vibe coding; it’s about literal programming. It’s not about accessibility, it’s about security. It’s about cost-affordability,” he said. 

IBM also announced new capabilities with AgentOps and watsonx Orchestrate. The goal: turning observability into an active improvement loop. IBM also released an extension of purpose-built AI agents to IBM Z, bringing agentic capabilities to the mainframe.

Beyond developers, AI capabilities have created a new level of expectations for enterprises. Organizations and businesses want automation that is proactive and contextual. But at the same time, they struggle with fragmented tools and an infrastructure that keeps getting more complex.

“Infrastructure today is highly fragmented,” said Armon Dadgar, HashiCorp CTO. “How do we solve that segmentation, make it accessible, both for our human operators and AI operators?”

For HashiCorp, a company acquired by IBM, the answer could be Project infragraph, a unified knowledge graph for infrastructure that connects apps and application workloads with their related components. The tool aims to help teams automate, enforce governance and scale AI safely.

“We are going through a business transformation like we have never seen before,” said Nirmal. “And that can only happen if enterprises adopt generative AI.”

Nirmal emphasized that only 5% of enterprises really see the value from their AI investments. Yet, the appetite to build and ship is growing stronger. Take TechXchange: IBM’s developer conference grew from 2,000 attendees in its first edition, in 2023, to more than 10,000 this year.  

“There are 2 million models available on Hugging Face. How many of them are actually usable in enterprise?” said Nirmal. Enterprises need AI tools that meet their security, governance and explainability. “This is why infusion of generative AI in enterprise takes time.”