Village Global Podcast

The episode contains a handful of genuine practitioner insights - the MCP token-bloat problem, the trust-gradient approach to agent adoption, and the unproven-system vs. untrusted-system distinction - but they are buried under 20+ minutes of personal biography and generic future-of-work optimism that adds no operational value.

“if you're using four, in the worst case you're using 80,000 tokens and you haven't done anything yet”

“She didn't not trust the system. She didn't trust an unproven system.”

The MCP-was-never-designed-for-remote-deployment argument is genuinely contrarian and grounded in engineering reasoning, and the DNS analogy for free integrations is a fresh framing; but the bulk of the episode recycles standard AI-will-free-us-from-computers discourse without first-principles depth.

“a lot of company got forced into, into adopting to MCP in a wrong way”

“I want to make one like the DNS. You don't expect when you go to the Internet and you type www uh with one AI, you don't expect that the DNS company is going to charge you for the lookup of the DNS”

Moe Katib is a legitimate 20-year integration practitioner who has verifiably built and shipped real infrastructure, giving him credible domain authority; however, One is an early-stage startup without demonstrated scale, and he has not yet operated at the level that would warrant top-tier caliber marks.

“for the last 12 years I've been working on the integration problem”

“we would work on one integration, literally one integration that would take us eight months and sometime even 12 months. And we'd have eight engineers working on one problem”

There are usable data points - 330 integrations, 98% first-run success, 80,000-token MCP worst case, $29 second tier, 1M free requests, first 100 integrations hand-vetted - but no revenue figures, customer counts, or named enterprise customers, and competitor comparisons are deliberately vague.

“we're getting roughly, uh, 98% success at first run”

“I believe today I checked, I think we were at 330”

The host asks a few genuinely useful follow-ups ('How do you trust that the agent worked?', 'Walk us through that decision to open source') but largely allows long biographical tangents to run unchecked, offers encouragement rather than challenge ('So. Innovation.'), and never pushes back on unsubstantiated claims like the 98% success rate or the assertion that agent adoption is primarily a trust problem.

“So. Innovation.”

“How do you trust that the agent worked?”