There is no one reason for the quality issues. It's a thousand small decisions and problems that have compound against each other over decades, coupled with the sheer feature complexity+scope+impact and multiplied by the titanic scale and volume the platform handles.
Additionally, the engineering culture really prioritizes backwards compatibility for customers (for good reasons) which bleeds into all aspects of the platform/decisions in both good and bad ways - and means that the big and obvious step-change platform improvements that could be made internally to make things better are not really invested in, or are deemed to expensive.
It's still a great place to work, and I'm proud that my work is in some small way directly contributing to and helping billions of people's work lives but there's still a long road ahead to improving the customer experience of using the platform for both internal and external customers.
Outlook already provides me this, it's terrible at it since context is key and context is probably buried in several places it has access to and despite that access, it still falls flat.
As I build this out there's actually less and less AI in the product and more good old-fashioned UX, writing and data entry tools, and automations.
Some examples...
We're simply bringing a CRM CRUD form into an email thread, populated from email sender / domain, for the end user to review and submit.
You can add your own notes into a thread, and copy / paste from
Similarly good pre-defined templates with variables perform way better than AI generated drafts.
Context is indeed key. The person at their email inbox has most of the context in their head, they need good tools to organize that context down for their future self and their team. AI can help but its really about just building a great tool for the operator.
This happened ~2014/2015.
https://newsletter.pragmaticengineer.com/p/how-microsoft-doe...