Microsoft merges Copilot into one corporate product and cedes personal chatbots to OpenAI, Google, and Meta

Some of us still remember TayAI from 2016, and that was before the Attention is all you need paper even existed.

Not surprising at all. Their enterprise offering is barely acceptable and the product cadence keeps slipping. All they had to do was copy Claude Cowork the way OpenAI did.

The difference with Microsoft is that it holds the enterprise data under governance agreements, so if its AI bypasses the guards a company puts on O365 data, Microsoft is liable. With Cowork and OpenAI the liability sits with the employee misusing the data.

Product remains an unmitigated dumpster fire, though GitHub Copilot has actually gotten to a decent place. And both products will ship features called autopilot doing entirely different things.

From the outside it seems so obvious. I know change at scale is hard, but it is still surprising nobody has called for a culling or a shakeup.

And unrelated to Windows Autopilot, which is device deployment in Intune succeeding SCCM. The naming is mindboggling.

Companies were paying for more than 30 million Copilot subscriptions as of the end of June, while the M365 apps bundle holding the strongest tools has about 90 million paying users. Also, cancel a home 365 subscription and they offer you a cheaper tier without AI.

Now if only they would start paying me cash for training their AI on my open source code instead of force feeding me a Copilot grant.

It was foolish to think they had any consumer clout left. Shoving a thousand copies of an inaccurate, inconsistent product down everyone's throat will eventually be a case study in destroying a brand.

Microsoft does not care about consumer branding because the enterprise branding game is A1. This is exactly how they have operated for decades.

From where I sit Microsoft is far from destroyed, it actually looks stronger than 10 or 20 years ago. Most users simply do not care about quality, accuracy, or functionality.

Crazy thing is, if they had waited 2-3 years and kept Windows AI-free, then released it today with the latest models, it would land really well. Instead managers chased promotions and pushed a technology that was not ready.

My blood pressure rises every time I use Enterprise Copilot at work. It truncates message history heavily and forgets the last thing I told it in the chat, like Momento mixed with Waiting for Godot.

Same feeling here since I switched to New Outlook. It is full of weird subtle bugs they take years to fix, so I am not optimistic about Copilot given the vision and execution are both questionable.

Doing research work with it that outputs any kind of artifact is painful. It crashes, has to regenerate, and the output barely carries to Sharepoint, so I end up building PDFs and importing them by hand.

Code snippets, especially HTML, frequently do not render, or if they do they render as an empty box and I have to switch each one to show code mode manually. Trillion dollar industry btw.

Not my experience with the premium license. The web chat with 5.6 thinking is fine, just a weaker interface than Codex or Claude Code, but the copilots embedded in specific Microsoft products are horrible and I consider them effectively worthless.

I am the exact target: Windows user, M365 subscriber, heavy VS Code and GitHub Copilot user, and basically every AI feature Microsoft ships is unusable garbage with the same models I find productive in other tools.

By the numbers the so-called AI doubters are most people. Heavy AI users are in the minority and the gap is only growing.

I think Microsoft makes far more sense as an agent orchestration layer over its OS and productivity software than as a competitor to the frontier model providers.