{
  "id": 4196392,
  "title": "The 3 roles AI agents play in your developer platform",
  "url": "https://urgent.news/2026/08/29/the-3-roles-ai-agents-play-in-your-developer-platform",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-29T13:00:00.000Z",
  "source": {
    "name": "The New Stack",
    "slug": "the-new-stack",
    "url": "https://thenewstack.io/ai-agent-platform-roles/"
  },
  "original_language": "en",
  "account": "Engineering organizations are racing to incorporate agentic AI into their developer platforms, seeking to boost productivity through AI agents. However, the role of these agents varies among teams. After speaking with hundreds of customers, three distinct roles have emerged.\n\nRole 1: AI agents as platform consumers are essentially users of the platform. They utilize the platform to gather context and perform actions. For instance, an engineer might ask Claude Code to add an endpoint to a payment service. Before writing any code, the agent retrieves relevant information from the platform, like the service owner, dependencies, and required standards. It then creates a preview environment, runs tests, and ensures the task is completed correctly. However, to ensure accurate context, the platform must provide a single governed source of truth - a context lake.\n\nRole 2: AI agents as internal platform components work within workflows triggered by events rather than individual requests. The agent operates within the platform, running alongside deterministic steps. For example, a nightly scan could flag vulnerable dependencies across multiple services. The remediation agent, pulled from a registry, runs once per service, allowing each owning team to receive an open PR awaiting review. This role requires an orchestration layer, a registry for agents, agent-specific identities for logging, and a human-in-the-loop step for significant risks.\n\nRole 3: AI as a resource with its own lifecycle, or AgenticOps, treats agents like any other resource like services, databases, or environments. The platform provisions, governs, and returns agents, much like a vending machine. An engineer could request an on-call triage agent, specify the model and tools needed, and the platform provisions everything required, complete with a well-governed agent. This role demands a self-service path to provision the runtime, issue identities and scoped credentials, wire in approved context, and register the agent for future reuse.",
  "summary": "Engineering organizations are trying to deliver as fast as technology allows, bringing agentic AI into their developer platforms and working The post The 3 roles AI agents play in your developer platform appeared first on The New Stack .",
  "key_points": [
    "AI agents serve as platform consumers, gathering context and performing actions.",
    "AI agents function as internal platform components, operating within workflows.",
    "AI agents can be treated as resources with their own lifecycle, akin to AgenticOps."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}