{
  "id": 7519412,
  "title": "Bring your own AI: the three ways to connect an assistant to a Birkly site",
  "url": "https://urgent.news/2026/09/15/bring-your-own-ai-the-three-ways-to-connect-an-assistant-to-a-birkly",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-15T09:28:00.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/rakete/bring-your-own-ai-the-three-ways-to-connect-an-assistant-to-a-birkly-site-41h2"
  },
  "original_language": "en",
  "account": "Birkly is an open, AI-native content management platform that allows users to build and manage digital experiences without vendor lock-in or complex technology stacks. One of Birkly's unique features is its approach to AI connections, treating them the same as human users in terms of roles and permissions. There are three ways to connect an AI to a Birkly site: personal connector, service agent, and built-in assistant.\n\n1. Personal connector: The AI acts as the logged-in user. This is the default option, with a single MCP URL for the site. Users paste this URL into their AI client and approve a consent screen, granting the AI the same permissions as the logged-in user. The AI's role is determined by the user's account permissions, with no separate AI role to manage.\n\n2. Service agent: This method assigns a fixed role to the AI, not tied to a specific person. Admins create service agents, assign them a fixed role (e.g., Editor), and decide which users can see them. Each service agent has its own dedicated MCP URL, allowing for controlled access without giving users full account access.\n\n3. Built-in assistant: This option uses an API key to connect directly to a user's AI model provider (e.g., OpenAI, Anthropic, Google). The AI acts within the admin panel, with an optional \"Require approval\" setting that requires human confirmation for write actions. This method allows users to use their preferred model provider without setting up a separate AI app.\n\nAll three methods share the same underlying permission model, with the only difference being the identity borrowed for the AI connection (person, service agent, or built-in assistant). Two site-wide settings further protect the system: the AI & MCP tools switch and the Plugin Security Denylist. The tools switch has three states - normal, read-only chat, and disabled - which override the AI's role permissions. The denylist prevents specific plugins from being accessed, regardless of the AI's permissions. Thus, Birkly's approach allows users to bring their own AI client, model provider, and identity system while maintaining tight control over permissions and security.",
  "summary": "Quick bit of context first, since this is a small project and there's a decent chance you're reading this without knowing what Birkly even is. Birkly is an open, AI-native content management platform that enables everyone to build, manage and operate digital experiences without vendor lock-in, unnecessary software complexity or dependency-heavy technology stacks. You can read more about it on…",
  "key_points": [],
  "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."
}