AI Desktop Wars: Who Gets The Crown?
Reflecting on May. Feels like another big shift in the landscape. Open I/O Buildathon (OpenAI + Google + Microsoft) has revealed a new battleground in the AI wars, the computer desktop and which AI Agents will control it. The big players know automating the browser is not enough and have said as much at their respective chest thumping events as they look at the full OS layer. If the browser was enough, we'd all be using Netbooks today 😅
The pieces are being put in place: desktop clients, multi-modal understanding, agents, and more. Is the ghost in your machine going to be a PC or Justin Long, or is it the year of the Linux desktop? Recall was likely shipped too early and is going to slow down Microsoft's work here, but we'll see, there are many ways to add value on the desktop with AI Agents without the always seeing eye.
OpenAI’s native desktop app + real time screensharing is going to evolve into acting, not just explaining. Your (definitely not ScarJo) AI friend will sit on the desktop and take over your mouse and keyboard if you let it quite soon, though it may be a Mac only product
Microsoft wants the same with Copilot(s), attempting a blockade. Clippy may rise from the dead after all spawning a Church of AI schism. The plugins approach is smart, but their renaming of everything to Copilot starts to have an IBM Watson problem. Even BillG is hinting at where this is all going.
Apple sits in a fog of war, but is alliancing for the intelligence to drive their systems with OpenAI. They have the right building blocks, but will probably go mobile-first and bring it back to the desktop later while OpenAI holds the strings.
Google, for their part, has the right multimodal model focus, but Android isn’t the play and Chromebooks are, well, Chromebooks. Still, their work in a new AI Workspace + Browser automation and new Agents tools could side step those issues, but I see them more as tools to use.
Even Nvidia has entered the fray this week (when their CEO isn't busy cosplaying a rockstar and signing "shirts") with their gaming assistant that has evolved from 2017 April Fools joke to a real live demo.
The winner will build a platform for devs to customize their own Agents/Assistants and embrace OSS. Your own fleet of assistants and agents should be personal, know your tools, abilities, and not be filtered heavily by committees.
There's still hope for an unknown startup to pull the sword and take their place 😉