Deployed ab345293 to develop in public with MkDocs 1.6.1 and mike 2.2.0

<h1 id="osl-ai">OSL AI</h1>

<figure>

  <a class="glightbox" data-type="image" data-width="100%" data-height="auto" href="../images/ai/osl_ai.png" data-desc-position="bottom"><img src="../images/ai/osl_ai.png" alt="OSL AI architecture"></a>

  <figcaption>Figure 1. OSL AI architecture</figcaption>

</figure>

<p>OpenSlice introduces a dedicated AI layer that brings conversational and agentic capabilities on top of the existing NaaS and Service Orchestration functionalities. The goal is to let stakeholders, like operators, developers, and end customers interact with OpenSlice through natural language, and to let OpenSlice itself use AI agents to support orchestration decisions.</p>

<p>The OSL AI layer is built around four concepts:</p>

<ul>

<li><strong>Agentic Orchestration</strong>: the ability for the Service Orchestrator to delegate complex orchestration decisions to cognitive agents <em>(Beta)</em></li>

</ul>

<p>Each of these is described below.</p>

<figure>

  <a class="glightbox" data-type="image" data-width="100%" data-height="auto" href="../images/ai/osl_ai.png" data-desc-position="bottom"><img src="../images/ai/osl_ai.png" alt="OSL AI architecture"></a>

  <figcaption>Figure 1. OSL AI architecture</figcaption>

</figure>

<h2 id="mcp-server">MCP Server</h2>

<p>The OpenSlice MCP Server implements a <a href="https://modelcontextprotocol.io/">Model Context Protocol</a> endpoint on top of OpenSlice's existing TMF-based service and product management stack. It is a standalone component that translates MCP tool invocations into calls against OpenSlice's internal catalog, ordering, and inventory subsystems, and returns the results in the structured format MCP clients expect.</p>

<p>Functionally, the exposed tool set is organized around six operational domains:</p>