EUC World: Independence 2024 — October 22
Brian gave the opening keynote at EUC World: Independence in Silver Spring, Maryland — the same building where BriForum, the community conference he co-founded, held its very first event 20 years earlier. He used the anniversary as the frame for a tour through two decades of end-user computing, arriving at a single argument: the Windows desktop, not some new AI-native platform, is about to become the API that AI agents plug into.
Twenty years of "faster horses"
Brian traced his own career from PC repair in 1994 through Systems Management Server, Citrix, VMware, VDI, mobile device management, and the consumerization of IT — a pattern he frames as each new tool being a faster horse for the same underlying job, not a reinvention of it. Citrix's original 1998 pitch was MAPS — Management, Access, Performance, Security — sold purely because it was a better way to run existing Windows applications, not because of any grander vision. VMware, then VDI, followed the same arc: initial skepticism, slow proof, eventual default.
Consumerization, and why Unified Endpoint Management never fully worked
The iPhone beat the BlackBerry in the enterprise not because of security or manageability, but because executives wanted Netflix and app vendors wouldn't build for BlackBerry's store — a reminder that employee preference, not IT strategy, has repeatedly been the actual force behind platform shifts. The industry's later attempt to unify management of desktops, mobile, and physical devices under a single "UEM" platform never fully delivered, Brian argued, because real organizations have too many overlapping use cases, regulations, and business units for one product to cover cleanly — what actually matters to a CIO is security, visibility, and whether employees can do their jobs, not which vendor's agent is installed.
"Windows is dead" has been wrong for 20 years running
Brian ran through the long list of technologies each declared to be the thing that would finally kill Windows — ASP.NET, Java, AJAX, HTML5, VDI, Chromebooks — none of which did. His point isn't nostalgia: it's that the desktop survives because it's a genuinely durable framework — a security container, an audit point, a place where applications already interoperate with each other — not just a legacy habit.
Microsoft's own roadmap is doing the arguing for him
Brian walked through a fast-moving sequence of real Microsoft and industry moves from the months leading up to the talk: Recall (continuous screenshotting, now framed as a stepping-stone to genuine intent-understanding, not just searchable screenshots); Copilot Vision (co-browsing the web with an AI that can see what you see); the Windows Agent Arena benchmark, where Microsoft defined ~150 concrete desktop tasks for AI agents and the leading agent scored 17% at the time of the talk; and a same-week report that Microsoft planned to ship Copilot agents capable of handling employee tasks within weeks. None of this was speculative — it was actively shipping.
The desktop is the API
The core argument: a company spends decades building the "perfect environment" for employee productivity — every internal app, every integration, every security control, already wired into the desktop. Rather than building fresh AI-native integrations for every one of those systems from scratch, point the AI agent at the desktop itself, the same way a human employee already uses it. "The desktop already has my security context, it has my audit trails, it's managed, I can move it around... the desktop is the API for all of your AI agents. I don't believe we need to reinvent this from scratch."
Putting that desktop in the cloud (VDI) makes it stronger, not weaker, for this purpose — the pixel stream already flowing to a human user can just as easily flow to an AI agent, with the exact same audit trail, compliance controls, and separation between work and personal use that VDI already provides.
His close, delivered as a talk he says he genuinely never expected to give: "2025 is going to be the year of VDI because of AI."
This is a summary. The full talk transcript lives in Brian's second brain: read it on GitHub.