shouche

We are moving beyond running local MCPs and are trying to map out a broader strategy - for internal as well as external consumption. I am curious how others are balancing business vision with technical reality.

Specifically:

* Commercialization & ROI: Executives view MCP servers as a new revenue stream (upselling like premium APIs). Is this realistic, or do MCPs eventually replace high-level APIs for agentic apps? How are you pricing this—fixed, or usage-based on "tool calls"?

* Infrastructure & Deployment: Are you rationalizing multiple REST/GraphQL endpoints into singular "task-based" MCP endpoints, or just wrapping existing APIs 1:1? How are you managing centralized discovery and rate limits across the org?

* Security & Governance: What do your "Human-in-the-loop" checkpoints look like for sensitive write actions? Are you using centralized MCP Gateways for secret management and audit logging?

* Product Development: How are non-tech product owners defining requirements? Have you found or built a "template" for spec-ing an MCP server that goes beyond a basic API spec?

In what ways wrapper apps built around LLMs such as GPTs or offer RAG based chatbots can provide significant value to a user?

For e.g., xplained.vercel.app that I made as a perplexity.ai styled clone that does RAG on web search.

Share your success stories and lessons learned, especially from a B2B perspective. Any tools that you anticipate will make a big impact in 2024?