He wrote the code that keeps Linux drivers alive through kernel upgrades. He helped build a CRM that Salesforce bought for $390M. He co-founded a calendar company that rescheduled 23 million meetings. Now he's designing how AI agents talk to each other - and to us.
The thing about Gary Lerhaupt is that he keeps building things that matter long after anyone expected them to. DKMS - the Linux kernel module support framework he co-created at Dell in 2001 - is still shipping inside Ubuntu, Fedora, and Debian today. More than two decades later. He wrote it in his twenties. Most engineers would consider that a career-defining contribution and stop there. Lerhaupt used it as a runway.
He went to Stanford for his master's in computer science and launched a BitTorrent media marketplace called Prodigem - essentially trying to build legal streaming distribution in 2005, a full decade before streaming was mainstream. Creative Commons spotlighted it. MoveDigital acquired it. He moved on.
He surfaced at Ludic Labs, which got absorbed into Groupon. Then he built HackerTable - a last-minute reservation system for the kind of restaurants where you need a reservation six months out. San Francisco magazine called it one of the coolest websites around. He moved on from that too.
Then he became founding engineer at RelateIQ. He rose to VP of Engineering. In 2014, Salesforce bought RelateIQ for approximately $390 million. That's when most engineers retire quietly into vest-and-relax mode. Lerhaupt co-founded Clockwise instead.
It won't be long before there are entirely new types of jobs out there where people use agents to do work in powerful and streamlined ways. The age of the Agent Orchestrator job description is upon us.
- Gary Lerhaupt, SalesforceClockwise was a different kind of bet - not a flashy consumer app or a moonshot hardware play, but a genuinely useful enterprise tool. An AI calendar that understood context. It learned your meeting patterns, blocked focus time, rescheduled low-priority events automatically, and worked across the entire team. It got used by Uber, Netflix, and Atlassian. Salesforce Ventures backed it. By the time it shut down in March 2026, it had created 8 million hours of uninterrupted focus time and rescheduled 23 million meetings. For a calendar app, those are extraordinary numbers.
Lerhaupt left Clockwise in 2025 with a LinkedIn post that said he was going "into the great wide open" - deliberately vague, clearly deliberate. A few months later he appeared at Salesforce with a VP title and a mandate to build Agent Interoperability and Orchestration for Agentforce. And then, in a twist he described as something "only Silicon Valley could write," he brought the entire Clockwise engineering team with him. Salesforce acquihired them. The product shut down. The team shipped out to build the infrastructure for the next era of enterprise AI.
That's Lerhaupt's pattern. He doesn't bounce between roles looking for the next bigger title. He follows the hardest unsolved problem of the moment - and tends to show up a few years before everyone else realizes it's the problem worth solving.
DKMS - Dynamic Kernel Module Support - is not glamorous. It's infrastructure-level plumbing. But it solved a real problem that drove Linux sysadmins into a quiet rage: every time you updated the kernel, your device drivers broke. You had to recompile everything by hand.
Lerhaupt, working on Dell's Linux team in the early 2000s, built a framework that automated this. When the kernel updates, DKMS rebuilds your modules automatically. It's the kind of thing nobody thinks about because it just works.
He wrote about it in Linux Journal in September 2003. The article is still indexed. The code is still running. That's what 20+ years of quietly correct infrastructure looks like.
The underlying philosophy - make complex systems composable and self-healing - shows up in everything he's built since. Clockwise made calendar management self-organizing. Agentforce's Agent Card makes AI agent capabilities self-describing. Same instinct. Different decade.
The enterprise AI problem isn't building smarter agents. It's making agents that can collaborate. Lerhaupt's Agentforce work is about exactly this - the protocols, standards, and infrastructure that let AI agents from different vendors work together without friction.
For agents to work together regardless of where a work request originates... a standard for doing this is needed.
- Gary Lerhaupt, on the A2A Protocol"In a twist maybe only Silicon Valley could write, this crew is joining Salesforce... I couldn't be more excited to build the future of AI alongside them again!"
On the Clockwise acquihire, LinkedIn 2026"Enterprises want controlled and simplified workflows that deliver repeatable value. And yet they don't want to be stuck in a silo."
Unite.AI, October 2024"When you are doing complex work, it's amazing to get to full automation, but it starts on the backbone of involving the user."
On human-in-the-loop AI"Your agents are in a black box... now you have the capability to see the full detail, see the underlying session trace."
On Agentforce observabilityThere's something almost novelistic about the arc. Lerhaupt first entered the Salesforce ecosystem in 2014, when RelateIQ was acquired. He stayed through the rebranding to SalesforceIQ, then left in 2016 to build Clockwise - with former RelateIQ colleagues, funded in part by Salesforce Ventures.
So Salesforce's own venture arm was backing the company its former VP of Engineering built after leaving. That's the kind of tangled-but-coherent relationship that Silicon Valley specializes in.
When Lerhaupt left Clockwise in 2025, the LinkedIn post was characteristically cryptic: "into the great wide open." No announcement. No press release. People who knew him probably assumed another startup. He surfaced at Salesforce with a VP title and a charter to build the infrastructure layer for enterprise AI agent collaboration.
And then the full circle completed: in March 2026, Salesforce acquihired the entire Clockwise engineering team. The Clockwise product shut down on March 27, 2026 - no IP transferred, no customer data moved, just the engineers walking through a different door. The team that built an AI calendar for Uber and Netflix is now building the protocols that will let AI agents work together across the enterprise. Same people. Bigger canvas.
The future of enterprise AI lies in seamless agent-to-agent collaboration.
- Gary Lerhaupt