Why AI Agents Cant Be Trusted Yet (And How to Fix It) | Moe Katib (One) @villageglobalvc
Why AI Agents Cant Be Trusted Yet (And How to Fix It) | Moe Katib (One)  @villageglobalvc
Uploaded June 2026 | Updated September 2026, 2 weeks ago
Moe Katib is the founder and CEO of One (withone.ai), an enterprise-level infrastructure platform that gives AI agents authenticated, reliable access to hundreds of software applications and verified actions. Before One, he spent 12 years solving enterprise integration problems, building the knowledge base that became the foundation for what he's building today.

Village Global GP Anne Dwane sits down with Moe to trace the story behind One: from growing up in Damascus selling packaged chicken out of his uncle's factory, to building water-pump automation devices growing up, to immigrating to Canada after his invention was stolen by a government-connected company. They cover what a decade of enterprise integration work taught him about agent infrastructure, why he made the counterintuitive decision to open source all the integrations he had assembled, and why trust, not speed or scale, is the real unlock for agentic AI adoption.

Moe also shares his own experience running his email entirely through an agent, including the moment his AI sent a rude message to a major VC on his behalf, and what that revealed about the identity problem at the heart of agentic software.

Timestamps:
00:00 Cold open
01:42 The stories that shaped Moe as a founder
01:57 Growing up in Damascus in an entrepreneurial family
04:38 The packaged chicken business: spotting his first product opportunity
08:05 The water pump automation device he built growing up
12:21 The woman who wouldn't use the new system, and what it taught him
14:52 Why trust, not features, drives software adoption
15:59 His vision for the future of work: away from the computer
19:49 From school software to enterprise integrations: the arc to One
22:34 How a 2024 experiment with Gmail convinced Moe agents were ready
23:30 How Moe defines agentic software today
24:21 Why agent adoption requires human trust above all else
25:44 The credit card analogy: what agents need to feel like
27:16 Introducing One: what the company actually does
29:30 The scenario generator: how One stress-tests every integration
31:49 Who is the ideal user for One?
37:17 The open source decision: convincing his team
40:50 Why users must own their own connections
42:33 How One will stay important as the agent landscape evolves
47:15 Moe's personal agent setup: no email login in months
52:04 The moment his AI sent a rude message to a VC
53:44 Should an agent have its own identity?
55:55 The elevator analogy: it just has to work
56:10 A real use case: managing an entire business finances agentically
58:38 MCP vs. integration: what's the actual difference?
01:01:24 The agent landscape: what's most promising right now
01:03:21 Final message: you are not going to lose your job
Why AI Agents Cant Be Trusted Yet (And How to Fix It) | Moe Katib (One)Meetring Advice From A Successful Entrepreneur And InvestorMark Pincus on the Contradiction Every Founder Hears Before They WinJeff Cobourn on Why the Gap Between Your Best and Worst Performers Is About to WidenWhy P-1 AI Charges a Salary Instead of a SubscriptionThe Consumer AI Strategy That Sets Meta Apart From Anthropic and CursorThe Email an AI Sent on Moe Katibs Behalf That Made Him CringeHow a Banner Next to a Ramen Shop Went Viral and Got Alma Real CustomersWhy Moe Katib Says Switching Away From Claude Feels Like Leaving a FriendWhy the Best Product Makers Are the Best CopiersThe Artistic Process Jade Q. Wang Says Looks a Lot Like Finding a Startup IdeaWhy Reid Hoffman Says AI Will Unlock Learning Potential in a Way Humans Never Could
Village Global |

Why AI Agents Can't Be Trusted Yet (And How to Fix It) | Moe Katib (One)

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER