Amazon Quick Desktop Is Now Generally Available: Pricing, Privacy, and Agent Controls

*This research-based analysis covers AWS's September 2026 general-availability announcement, documentation, security architecture, and published pricing. DiscoverAI has not independently tested the desktop client; product and privacy descriptions attributed to AWS remain provider claims.*

Amazon Quick is now generally available as a desktop application for macOS and Windows. It can connect to local folders, email, calendars, messaging, business apps, browser automation, and local MCP servers, while agents continue cloud work after the computer closes. AWS also added a mobile activity feed that prioritizes items needing human decisions while agents handle routine work.

Quick has a free individual plan. Plus costs $20 per user per month billed annually or $25 monthly, while Max costs $100 annually or $125 monthly. Enterprise-oriented Professional and Enterprise plans are $20 and $40 per user per month respectively, each with a $250 monthly infrastructure fee per account; allowances and overages differ by plan.

The app moved from preview to general availability and expanded to seven AWS regions: Northern Virginia, Oregon, Sydney, Tokyo, Ireland, London, and Frankfurt. AWS says it maintains production stability and applicable service-level commitments while keeping conversations and agents synchronized across desktop and mobile.

Desktop access changes the product's authority. Quick can work with local documents, spreadsheets, presentations, images, PDFs, and code; surface notifications; operate browser or file tools; and connect to local coding agents through MCP. A task can continue in the cloud after a laptop closes, so “background” should not be interpreted as local-only processing.

Free includes the desktop app, individual agents, research, deliverable creation, Flows, app creation, and connectors, with published capacity limits. Plus adds team spaces, administration, extensions, and 10 GB of pooled index storage per seat. Max includes five times more usage capacity and 50 GB per seat.

For AWS-managed organizations, Professional is $20 per user monthly plus a $250 account infrastructure fee and includes eight agent hours and 25 GB per user. Enterprise is $40 per user monthly plus the infrastructure fee, includes 18 agent hours and 50 GB, and adds complex multi-step automation. AWS lists Enterprise overage at $3 per agent hour and $5 per GB of index storage; Professional currently cannot enable overages.

The practical unit is an accepted completed workflow, not a seat. Teams should measure background-agent hours, index growth, failed tasks, reviewer time, and downstream action errors before extrapolating a pilot budget.

Not necessarily. AWS says the desktop app stores local configuration, granted-folder settings, locally configured MCP servers, and authentication tokens on the computer. Conversations, memory, knowledge graph, agent-created Library files, and searchable indexes live in the user's cloud Quick account.

Adding a folder grants agent access, while separate options can upload full file contents for search or extract people, projects, and dates into the knowledge graph. AWS says customer conversations, files, and personal context are not used to train its AI models. Teams should still classify folders before connection, exclude secrets and regulated records, choose read-only access where possible, and test deletion and retention behavior under their own account type.

System tools, third-party connectors, and local tools can be set to Always Allow, Ask Each Time, or Always Deny, with granular read and write operations. Quick uses OAuth for connected services and AWS says it does not store third-party passwords. Private Mode prevents a conversation from being saved, learned from, or shown in Recents.

Those controls reduce exposure but do not prove an agent understands intent. “Always Allow” can turn one ambiguous prompt or hostile document into repeated action. Start with Ask Each Time for writes, outbound messages, browser automation, downloads, and code execution; use dedicated low-privilege accounts; and inspect the session activity before broadening authority.

It is most relevant to teams already using AWS or to knowledge workers who want one agent surface across documents, communications, analytics, and automation. It is a weaker fit when policy forbids cloud indexing, supported regions do not meet residency needs, local folders cannot be cleanly separated, or simpler assistants already solve the job without persistent context.

Pilot with one folder, one connector, and one reversible workflow. Seed stale files, conflicting instructions, a malicious prompt inside a document, and an action that should require approval. Score source selection, task completion, permission behavior, recovery, cloud data created, agent hours, and human correction.

Amazon Quick desktop makes enterprise agents more accessible by putting local files, connected work, background execution, and a persistent knowledge graph in a native app. The free plan makes the interface easy to test, while AWS governance and regional deployment options give larger organizations a credible adoption path.

The same convenience enlarges the trust boundary. Quick runs in the cloud, some indexing choices upload local content, and agents can retain context or act after the machine closes. Treat permissions, indexing, background work, cost ceilings, and deletion as acceptance tests—not settings to revisit after rollout.