When does MCP make sense?

Bill Doerrfeld | September 11, 2025

When does using MCP make sense? My latest for The New Stack cuts through the hype to answer this loaded question head-on.


The general consensus is that MCP is a "force multiplier" for indeterministic workflows, novel integrations, and giving AI coding agents context on the fly.


But for more predictable automation, typical CLI commands or API integrations are going to be more reliable in the long run. Anything else is probably overengineering.


Also, one key best practice: You shouldn't be mapping APIs one-to-one — MCP tools should model user intents and consider natural language.


My piece goes a bit deeper and looks at some emerging MCP server success stories at GitHub, Atlassian, Red Hat, and beyond.


This is definitely an evolving area, but I hope this piece tries to make sense of when it's best to put MCP to use!

Read: When Is MCP Actually Worth It?

Other Blog Posts

KubeCon + CloudNativeCon North America 2026 keynote stage with the
By Bill Doerrfeld • November 10, 2026
At KubeCon NA 2026, the opening keynotes made it clear: AI inference is the new cloud native production workload. Here's what that means for Kubernetes.
Agentics Day: MCP + Agents North America title slide on the screen at KubeCon NA 2026
By Bill Doerrfeld • November 10, 2026
Agents are getting production tested. I explore some of the top lessons learned at KubeCon NA 2026's co-located event, Agentics Day: MCP + Agents.
KubeCon + CloudNativeCon North America 2026 graphic: November 9-12, Salt Lake City, Utah
By Bill Doerrfeld • November 6, 2026
I'll be at KubeCon + CloudNativeCon NA 2026 in Salt Lake City, Nov 9–12, scouting stories on agentic AI, cloud native and APIs. Reach out if you'd like to connect!
AI Pen-Testing Conference graphic from AI Security University
By Bill Doerrfeld • November 4, 2026
Another virtual conference is in the bag, and the community really showed up to talk AI penetration testing.
By Bill Doerrfeld • October 30, 2026
My eight-part Road to KubeCon column with The New Stack is done. Here are some of the key takeaways.
How AI Agents Change API Interaction opening slide doerrfeld bill 2026 nordic apis
By Bill Doerrfeld • October 18, 2026
Here are my slides from my opening keynote at Nordic APIs Summit 2026, exploring how AI agents are changing API interaction patterns.
By Bill Doerrfeld • October 17, 2026
My latest research for AI Security University looks at AI tools and resources for pen-testers, red teamers, and bug bounty hunters.
By Bill Doerrfeld • October 16, 2026
My commissioned feature for Arthur D. Little's PRISM publication analyzes the big shifts underway now that generating code is becoming trivial.
By Bill Doerrfeld • October 11, 2026
Was happy to participate as a speaker and moderator at SecureMaine on October 8th, 2026.
By Bill Doerrfeld • October 5, 2026
My latest for LeadDev interviews a Meta engineer who built an agentic system to optimize their capacity planning, leading to real power use savings.