IBM announced a new set of agent governance capabilities at its TechXchange 2026 conference in Atlanta on Tuesday, October 27, 2026, built around a simple idea: no AI agent should act inside a business without a verified identity, a defined job, and a way to stop it instantly. The package combines HashiCorp Vault’s credential controls with a new Agent Identity record, plus discovery, containment, and evaluation tools in watsonx Orchestrate, and it arrives as enterprises deploy agents far faster than they can account for them.
An Identity Record For Every Agent
HashiCorp Vault already manages the secrets and credentials that applications and services use to reach business systems, through centrally managed secrets, automated rotation, and dynamically generated credentials that expire or can be revoked. IBM is now extending those capabilities to AI agents, which it describes as a growing class of non-human identity alongside applications and services.
When used with Vault, Agent Identity establishes the authoritative record for each agent, including who owns it, its lifecycle status, and the authority it has been granted. That record then feeds Vault’s credential controls, enforcing least-privilege access and enabling real-time revocation when an agent attempts an unauthorized action. If an agent tries to retrieve information or perform an action outside its approved permissions, access can be denied and the session revoked immediately, rather than discovered in an audit weeks later.
Vault 2.2, generally available as part of the announcement, gives agents their own secured accounts, automatically renews their credentials, and incorporates post-quantum encryption for quantum-era threats. A new Secure Hub, in public preview, gives companies centralized visibility across their Vault environments, so they can see operational health, usage, and security resources in one place.
Finding, Containing, And Testing Agents
Identity alone does not solve the sprawl problem, because many enterprises cannot even list the agents already running inside them. IBM watsonx Orchestrate is designed to work across the AI ecosystem rather than inside a single vendor’s platform, giving organizations one place to discover, secure, and evaluate agents wherever they were built.
Its Discover capability, in tech preview, connects to platforms such as Amazon Bedrock, Microsoft Foundry, and Google Gemini to find agents built in third-party environments, and extends to models and MCP servers. Its Contain capability, also in tech preview, runs agents in isolated sandboxes so a rogue agent cannot interfere with others, and gives teams a kill switch to shut down any agent outright.
The Evaluate capability, generally available, lets teams test the quality, accuracy, and compliance of agents, models, and MCP servers before deployment, then monitor them in production. Built-in cost tracking shows each agent’s model and token use, so teams can spot what is driving spend and fix it, which matters when an agent can quietly burn budget at machine speed.
Connecting Agents To The Business
Once agents are authorized to use a company’s APIs and tools, they need to know what already exists, how systems connect, and what their actions could affect. IBM Integration Intelligence Hub, in tech preview, provides that knowledge through a Unified Context Engine, so agents can fix problems like stuck messages or deploy an API using what is already in place, with skills, approval policies, and audit controls keeping those actions in check.
IBM also announced that Agent Identity and HashiCorp Vault integrate with NVIDIA OpenShell secure runtime software, pairing a protected environment for agents with a verified identity and short-lived credentials for every agent, so each action can be traced, limited, and reviewed. John Fanelli, NVIDIA’s vice president of enterprise software, said the combination helps enterprises scale agents with secure execution, verified identities, and tightly scoped access.
The numbers behind the announcement explain the urgency. According to Gartner, cited by IBM, the average Fortune 500 enterprise will use more than 150,000 AI agents by 2028, yet only 13 percent of organizations believe they have the right agent governance in place. Persistent credentials, unclear ownership, and disconnected controls are the gaps IBM says its package targets.
Why It Matters
This matters because agent governance is becoming the enterprise AI bottleneck, not model quality. An agent with a long-lived, overprivileged credential is a security incident waiting for a prompt, and an agent nobody owns is unauditable by definition. IBM’s answer, identity plus short-lived credentials plus a kill switch, is the same discipline enterprises already apply to human and service accounts, finally extended to agents. It also puts IBM in direct conversation with other vendors racing to verify agent identity, with suites bringing ChatGPT into the enterprise, and with evidence that people already defer to agents most of the time, which makes the controls around them more urgent, not less.
Background
IBM TechXchange is the company’s annual conference for developers and technology practitioners, drawing nearly 6,500 attendees from 85 countries this year, and the event now incorporates HashiConf, bringing the HashiCorp community together with practitioners from IBM, Red Hat, and Confluent. Dinesh Nirmal, senior vice president of IBM Software, said the winners in enterprise AI are changing how their businesses run, not simply adding more AI to them.
Source: IBM announcement
0 Comments