BOSTON -- A generally available AI coding agent and previews that reveal more detail about IBM's plans to tie together its AI-driven automation tools topped updates for developers and platform engineers here at the IBM Think conference this week.
IBM Bob, previewed in October and shipped on April 28, replaces the previous Watsonx Code Assistant. It can generate application code similar to the previous agent but now also automates full software deployments by orchestrating other AI agents, connecting to Model Context Protocol (MCP) servers and using multiple IBM tool sets. IBM Bob can automatically route tasks among multiple large language models based on their suitability. A new IBM Bob Premium Package for Z in private preview will integrate with mainframe applications.
IBM Bob enters a market thoroughly saturated with AI coding agent alternatives from frontier model companies and enterprise IT vendors alike, including Anthropic Claude Code and Cowork, Microsoft GitHub Copilot and Factory, Amazon Q Developer and AWS Kiro, and Gemini CLI, Code Assist and Cloud Assist. Anthropic and AWS also offer tools specifically for mainframe application modernization and development.
IBM claims unique expertise given the central place IBM Z mainframe systems occupy in the market, however.
Rob Thomas
"COBOL is machine-level code, meaning that when you try to run it somewhere else, you normally give up a massive amount of fidelity and performance characteristics," said Rob Thomas, senior vice president of software and chief commercial officer at IBM during a press briefing this week. "We focused with Watsonx Code Assistant and now IBM Bob for Z on … scanning the code so we can actually see the different business objects and the business logic."
IBM Bob joins AI coding agent fray
IBM Bob fills a crucial gap in IBM's product portfolio and creates a new entry point into its underlying IT automation toolset for existing customers, but hasn't outpaced its many competitors with its specific features, said Jason Andersen, an analyst at Moor Insights & Strategy.
"They started to realize that as a vendor that provides back-end services, they [needed] something that was more specific to the user space, and going into it from an application development user persona made a lot of sense," Andersen said. "IBM does have a lot of what we'll call non-standard platforms and languages, so for them to have their own specific tool set, instead of building out a set of plugins … is going to give them some better integration opportunities and allow them to move a little faster."
Bradley Shimmin
Worries about the security of open source AI coding agents such as OpenClaw also make it important for IBM to offer its customers a way to create customized agentic workflows they design themselves, but with built-in enterprise governance, said Bradley Shimmin, an analyst at Futurum Group.
"Claude is cool if you write code, but it's built to accommodate the masses. You buy an IDE from JetBrains, and they're trying to be something to everybody. Same with VS Code from Microsoft," Shimmin said. "What we're coming to in the enterprise is a self-driven, self-determined, curated tool chain that is built by the people that are writing the code, those who are responsible for the outcome. They're not waiting on or even relying on MCP hubs or [things like] Claude Cowork."
IBM Concert, HCP preview Infragraph
Another set of preview-stage tools this week brings the joint HashiCorp-IBM knowledge graph project, Infragraph, to HashiCorp Cloud Platform (HCP) Terraform and IBM's Concert to broaden its AIOps features. IBM began talking about AIOps integration with HashiCorp through Concert two years ago, but the initial focus for the tool was on application security Thomas said.
"A year ago, we talked about Concert as a way to apply AI to identify security vulnerabilities in your landscape," Thomas said. "I would describe Infrafraph as a multi-cloud CMDB [configuration management database], a way look at all of your systems in real time … and then Concert is the AI that you apply on top of that … to automate and do management across hybrid environments."
The Infragraph integration will underpin a new, broader IBM Concert Platform that covers six areas of hybrid cloud management, according to an IBM blog post. These areas include:
Concert Observe for observability with integrations into IBM Instana and SevOne;
Concert Optimize with performance and cost optimization from Turbonomic;
Concert Operate for incident detection, investigation and response from IBM Cloud Pak for AlOps;
Concert Protect for Al-driven vulnerability and risk management;
Concert Resilience for reliability insights and proactive operations;
Concert Workflows for low-code workflow orchestration and automation.
Concert Platform marks another necessary step for IBM to bring its various acquisitions together more cohesively, Andersen said.
"IBM is recognizing that they needed to start bringing those products more closely together, so customers could get better benefits," Andersen said.
In the end, Concert [will] be the glue that brings it all together.
Jason Andersen,Analyst, Moor Insights & Strategy
The likely next area of focus for IBM Concert will be resiliency, an increasingly important part of cybersecurity among enterprises, Andersen predicted.
"It's going to expand into an autonomous health overlay, or an engagement system over all these discrete services like Cloudability, Apptio, Instana, HashiCorp Vault, maybe even Red Hat products," he said. "In the end, Concert [will] be the glue that brings it all together."
IBM Concert also added a feature this week called Secure Coder, in public preview this week, which can automatically prioritize and remediate security vulnerabilities in code, operating systems, middleware, packages and images through IBM Bob and VS Code IDEs.
"Concert has gotten a lot of attention since the news on [Anthropic's] Mythos came out, because every customer is looking for, 'How do I solve problems with vulnerabilities that I'm not aware of?'" Thomas said.
More IBM AI updates at Think
IBM Bob and Concert updates were just two among a broad set of product news at Think, which also included the first integrations between IBM's recently-closed Confluent data pipeline acquisition and its Watsonx.data lakehouse. IBM's Sovereign Core hybrid cloud AI framework became generally available after two months in preview. And the Watsonx Orchestrate AI agent orchestration platform added a new AI agent catalog with support for third-party AI agents from ServiceNow, Salesforce and Adobe, among others.
Knowledge graphs are becoming a common means to ground autonomous agents in context on various vendors' platforms. Cross-domain agent orchestration that encompasses third-party agents and tools has also become a common selling point among enterprise vendors.
Those trends reflect one of the core problems with enterprise AI agent adoption: a widening AI skills gap, according to Shimmin.
In two Futurum Group Data Intelligence, Analytics, and Infrastructure Decision Maker Surveys conducted in September and January, about half of more than 800 respondents identified agentic AI as a key objective. But in the January survey, the number reporting an AI skills gap as a top organizational challenge doubled, from 5% to 10%.
"What an integrated stack [such as Watsonx Orchestrate] does is convey performance gains and cost optimization," Shimmin said. "An integrated stack also leads toward what we're seeing with HashiCorp and IBM, where … even if it's opinionated … they're going to alleviate some of the heavy lifting [for customers]."
Beth Pariseau, senior news writer for Informa TechTarget, is an award-winning veteran of IT journalism. Have a tip? Email her or connect on LinkedIn.
Dig Deeper on Systems automation and orchestration