You must implement a system of governance that treats the AI agent with the same, or stricter, diligence you would apply to hiring a human employee.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

Gartner predicts 40% of agentic AI projects will be canceled by 2027, primarily due to inadequate guardrails rather than model limitations. Successful AI agent deployment requires treating them like human employees with strict governance: implementing technical constraints (data redaction, action limitation, output validation), using outcome-oriented testing with model-graded evaluations and golden datasets, deploying specialized agents coordinated by an orchestrator instead of monolithic systems, and following a phased rollout (shadow mode, human-in-the-loop, full automation). Even autonomous agents need ongoing oversight with clear managerial accountability to prevent costly failures like the Chevrolet chatbot that agreed to sell a $76,000 vehicle for $1.

Why 40% of AI Agents Might Fail (and How To Save Yours)

Guardrails: The AI Agent’s Job Description

Testing Agentic AI: Moving Beyond Unit Tests

How To Design AI Agents To Support Full Autonomy