Amazon this week developed Amazon Quick with a desktop app for Windows and Mac and Microsoft 365 extensions that put Quick in Outlook, Word and Teams, with PowerPoint and Excel extensions still in preview. The expansion arrives about six months after Quick’s October launch and signals an obvious admission. Quick needs to live where knowledge workers already live, and that place is Microsoft 365.
This is the structural problem that Amazon has spent more than a decade refusing to solve. Quick is the latest effort to sell workplace productivity software to knowledge workers, following WorkDocs, Amazon Chime and WorkMail. Three of these older products are now discontinued. WorkDocs ended support in April 2025. Chime was discontinued in February 2026. WorkMail ends support in March 2027. Retirements are accumulated in a 23-month window. Quick is technically different. The market problem it addresses is the same.
What Amazon sells
Quick is one teammate agent that combines research, business intelligence and workflow automation into a single workspace. Its components include Quick Index for unified enterprise search, Quick Search for autonomous multi-source analysis, Quick Vision for natural language analytics, and Quick Flows and Quick Automation for process orchestration. Quick builds on technology that Amazon previously shipped as Q Business and QuickSight, and existing Q Business customers can carry their indexes forward.
Quickly connect to Microsoft 365, Google Workspace, Slack, Salesforce and Zoom. It supports Model Context Protocol for integrating third-party agents such as Visier’s workforce analytics. Name AWS Vertiv, DXC Technology, 3M and Jabil as customers of Amazon as well communication it lists GoDaddy, AstraZeneca, BMW, Mondelez, NFL and Southwest Airlines among its current users. The potential set is real. Whether it is enough to overcome the distribution problem is a different question.
The Cowork market has already formed around Microsoft
Three “Cowork” agent products now compete for the desktop knowledge worker, and Amazon comes in last with the weakest distribution position. Anthropic has started Claude Cowork in January 2026 on macOS and brought it to Windows in February. Microsoft started Copilot Cowork through the Frontier program at the end of March. The Microsoft product operates on technology introduced by Microsoft from Anthropic and according to Microsoft Learn it’s available in the browser, in Outlook and Teams, and in the Microsoft 365 Copilot desktop app, with Work IQ pulling context from across Outlook, Teams, Excel, files, and meetings.
This partnership is the most important event in understanding Quick’s position in the market. Microsoft is now shipping Anthropic’s representative technology into applications where its enterprise base is already running. Amazon ships Quick as a separate destination with extensions to those same apps still in preview.
Microsoft’s adoption math shows how difficult it is to close the Surface gap even with a first launch. On him Q2 2026 earnings callMicrosoft reported 15 million paid Microsoft 365 Copilot seats against more than 450 million commercial Microsoft 365 subscribers. That’s about 3.3 percent paid penetration of the broader enterprise productivity software install base, after two years of distribution through the apps users open every morning. Amazon does not have comparable distribution and starts later.
Amazon’s pattern has not been solved
The shutdowns of WorkDocs, Chime and WorkMail were not technical failures. Each product required knowledge workers to ditch the tools they were already using and adopt an Amazon-branded alternative. None of them made it to scale. Amazon’s enterprise workforce reportedly runs on Microsoft 365, according to Business Insider’s report summarized by GeekWire and others, a procurement reality that has served as a quiet verdict on Amazon’s products.
Quick reframes the problem rather than solving it. Where WorkDocs used to compete with Microsoft and Google for document collaboration, Quick tries to work on top of the productivity suite through plugins, indexing and links. The bet is that the agent layer matters more than the application layer, and that an AWS-hosted agent that reads Microsoft, Google, and Slack is more valuable than a native Microsoft agent that reads Microsoft. Microsoft is testing this bet under home conditions with Anthropic’s technology already in the suite.
What Quick Has Going For It
The case for Quick is not zero. Q Business gave Amazon a business payment base prior to the launch of Quick, which is larger than WorkDocs ever had. The key relationships put AWS in front of buyers who already source AI infrastructure from the company. MCP support and a range of connection slots make Quick reliable in heterogeneous environments where Microsoft 365 is one of many systems rather than the center of gravity. Microsoft 365 extensions, when they come out of preview, could reduce surface space. There is a plausible version of the future for Quick where it becomes the multi-suite agent tier for organizations that aren’t fully committed to Microsoft or Google. This result is less than what AWS has positioned Quick to achieve.
Practical limitations
Three constraints will shape Quick’s trajectory. The PowerPoint and Excel extensions remain in preview, which means that Quick’s most important distribution channel is not yet generally available. The product was released against a market where the leading competitor uses the same representative technology that Amazon would have to license or develop. And the historical pattern, that Amazon’s productivity products fail not on capability but on adoption, remains unchallenged by anything Quick has shown.
There is also a model question. Microsoft brought Claude to Microsoft 365. Google ships Gemini natively in Workspace. Quick’s underlying model stack is not central to its public position, which in itself is a sign that Amazon leads with workflow rather than intelligence.
What it means for business
For technology leaders evaluating Quick, the curation question is not whether the product works. That’s where users will encounter it and what the active usage rate looks like in the sixth month. Junior pilot. Measure commission-to-active conversion against the Microsoft Copilot benchmark of 3.3 percent paid penetration. Ask vendors how their agent reaches users in Outlook and Teams without a context switch, and treat preview extensions as a roadmap rather than a current feature.
The deeper question is whether the agent layer is compressed into the application layer or remains a separate market. If Microsoft’s bet is correct, any meaningful agent capability ultimately lives inside the productivity suite that users are already paying for, and autonomous agent platforms become a category waiting to be absorbed. If Amazon’s bet is correct, the agent becomes the new functional layer and the suite becomes plumbed. Both bets can’t be right, and Amazon is playing the hardest with the weakest hand.
WorkDocs, Chime, and WorkMail didn’t fail because Amazon couldn’t build software. They failed because Amazon couldn’t change where the work was done. Until that changes, every Amazon productivity product is a bet against gravity.
