The reality of building on MCP: Is the friction worth it?

We’ve spent the last few months moving our Google Workspace tools over to the Model Context Protocol (MCP). While the potential for 'agentic' workflows is huge, but the friction of working on the edge of a new protocol is very real.

For those who haven't dived in yet, the biggest hurdle we found was the Remote vs. Local setup, rather than the protocol itself. Most tutorials focus on local command-line installs, but for a production-ready SaaS, you have to build for remote MCP servers. This requires a completely different approach to authentication and persistent state that the current docs don't fully cover yet.

We decided to take this 'hard path' because the shift from chatbots to execution layers feels inevitable. If the AI can't actually act on the data it sees, it’s just another source of noise.

If you’re experimenting with MCP, are you finding the 'Remote Server' implementation a barrier for your users? Or are you seeing a preference for local-first setups? I’m curious to hear from other makers.

We’re prepping to battle-test this approach in a public launch soon, and I'd love to get inputs from anyone else handling the implementation hurdles.