Publisher Guide: Should You Configure Your Own MCP Server or Issue a License?

As a publisher on Cashmere, you have two paths for making your content available to AI-powered applications:

1. Configure and operate your own MCP server — you are the one delivering the end-user experience.
2. Issue a content license to an application developer — they configure everything; your role is content supplier.

The right choice comes down to one question: Who is responsible for the end-user relationship?

---

Decision Tree

---

Path 1: Configure Your Own MCP Server

Choose this path when you are the one building and operating the application that end users interact with.

When you run your own MCP server you are responsible for:

• Setting up and maintaining the MCP server configuration
• Designing the tools, tool descriptions, and retrieval behavior
• Running evaluation tests to ensure quality and relevance
• Managing subscriptions or access for your end users
• Handling licensing of the content within your server

Use cases that fit Path 1

• You are launching a subscription AI product. For example, a medical publisher building a clinical-decision support tool that clinicians pay to access. You own the product, the billing, and the user experience.
• You are building an internal enterprise tool. Your organization wants a private, branded AI assistant powered by your proprietary content library. You control who gets access and how.
• You are a professional association offering a member benefit. You want members to search your journal archive through an AI interface you manage — not one run by a third party.
• You are a research platform aggregating multiple titles. You license content from several sources, bundle it under your brand, and sell subscriptions. The MCP server is the backbone of your product.

In all of these cases, you are the application developer. Follow the Developer Quickstart to get your API key, configure your MCP server, and connect it to your preferred AI client.

---

Path 2: Issue a License to an Application Developer

Choose this path when someone else is building the application — and your role is to supply high-quality, licensed content into their product.

When you issue a license:

• The developer (licensee) handles all MCP server configuration and tooling
• The developer manages their own end-user relationships and subscriptions
• You focus on content quality, collection organization, and license terms
• You control what rights the licensee receives (e.g., RAG only, or RAG + READ_RAG_SOURCE)
• You can approve licensees manually or enable auto-approval for public collections

Use cases that fit Path 2

• A SaaS company wants to enrich their AI assistant with your content. They're building a productivity app for lawyers and want to incorporate your legal reference titles. They configure the MCP server; you issue them a license with RAG rights.
• An edtech platform is integrating your textbooks. They have an existing student-facing AI tutor. They handle the infrastructure; you supply the content and set the MPT rate.
• A consulting firm wants your research reports for internal RAG pipelines. They are not building a product — they want your content feeding their private knowledge base. Issue a license; let them configure their own server.
• An aggregator is building a multi-publisher AI library. They bring together content from many publishers under one product. You are one content supplier among many — issue a license and let them handle the application layer.
• A developer is building a general-purpose AI tool and wants to offer your content as a premium add-on. They handle the product; you supply the content through a license.

To set up licensing, organize your content into Collections, set your MPT rate, and configure whether the collection is public (discoverable) or private (invite-only). You can control access rights using License Rights (RAG and/or READ_RAG_SOURCE).

---

Can You Do Both?

Yes. Many publishers act as both an application operator and a content licensor simultaneously. For example, you might run your own branded AI product (Path 1) and also license your content collection to third-party developers (Path 2). Cashmere supports both workflows from the same account.

---

Related Resources

• Developer Quickstart
• Creating Omnipub Collections
• Types of License Rights
• Cashmere MCP Tools
• Managing Licenses via API