Amazon Quick Launches With Free Tier, Desktop App, and MCP Server

Amazon launched Quick at AWS Summit NYC as a full replacement for Q Business, an agentic AI work assistant with a free tier that does not require an AWS account. The platform includes a desktop app that indexes local files without uploading them, native MCP server support for enterprise connectivity, and four pricing tiers starting with a free single-user option.

Amazon Quick arrived at AWS Summit NYC https://aws.amazon.com/events/summits/new-york/ today as the company’s full replacement for Q Business — and the most notable thing about it is what it doesn’t require: an AWS account. You can sign in with Google, Apple, or GitHub credentials and start using a genuine free tier that includes custom agents, Spaces, and knowledge bases. For a company that has historically used AWS account creation as the toll booth to every product, this is a meaningful shift. What Amazon Quick Actually Does Quick is an agentic AI assistant for work. That distinction matters because it’s not a chatbot with company data attached — it’s a five-module platform that can schedule meetings, generate visual assets, run background agents on a schedule, and connect to external tools through native MCP support. It replaces Amazon Q Business, which was a reactive interface: connect your data, ask questions. Quick takes actions on your behalf. The five modules cover chat and collaborative Spaces, custom agent building with a drag-and-drop Agent Builder and pre-made templates , knowledge bases, visual asset generation documents, presentations, infographics , and extensions and MCP integrations. Teams can build purpose-specific agents and share them across Spaces — which means a support team can build a ticketing agent and an analytics team can build a dashboard agent, both inside the same Quick workspace. The Desktop App Changes the Privacy Calculus The desktop app for macOS and Windows https://aws.amazon.com/about-aws/whats-new/2026/04/amazon-quick-macos-windows-preview/ — currently in Preview — addresses one of the main objections to cloud AI assistants: you have to upload everything. Quick’s desktop client reads and indexes local files without requiring upload. Your files stay on your machine. The app connects to your calendar, surfaces OS-level notifications for action items and conflicts, and runs background agents that monitor Slack, email, and calendar to surface what matters through a unified Activity feed. This is not the same as browser-based assistants that require you to paste in content or upload documents. If a file is in a folder you’ve granted access to, Quick can find it, summarize it, and act on it — without it leaving your machine. That’s a different security model than most enterprise AI tools are offering right now. MCP Server Support: The Enterprise Unlock Amazon Quick ships with a native MCP client. You can connect to any MCP server reachable over the public internet, but the enterprise-critical feature is VPC connectivity: connect to private MCP servers reachable from a VPC in your AWS account, which means internal tools, on-premises systems, and proprietary databases can talk to Quick without internet exposure. The connection can go through VPC peering, Transit Gateway, AWS Direct Connect, or VPN. Security uses PKCE with the S256 challenge method and Resource Indicators RFC 8707 to bind access tokens to specific MCP servers. That’s the protocol-level detail that regulated industries care about. AWS’s MCP integration guide https://aws.amazon.com/blogs/machine-learning/integrate-external-tools-with-amazon-quick-agents-using-model-context-protocol-mcp/ covers the full setup. New Relic and Moody’s already have live MCP integrations — New Relic for incident investigation and NRQL queries, Moody’s for ratings and research data on over 600 million entities. Pricing: Four Tiers, One of Which Is Actually Free The pricing structure replaces Q Business’s simpler model with four tiers: Free: Chat, Spaces, custom agents, knowledge bases — for a single user, indefinitely Plus: $20/user/month — higher limits, full integration access Professional: $20/user/month + $250/account/month infrastructure fee Enterprise: $40/user/month + $250/account/month infrastructure fee There’s also a 30-day free trial for up to 25 users at Professional tier with no infrastructure fee and no credit card required. After 30 days, accounts without payment revert to the free tier — not a hard cutoff. At $20/user, Quick Plus is $10/user cheaper than Microsoft 365 Copilot $30/user and $40/user cheaper than ChatGPT Enterprise $60/user . At 100 users, that’s $12,000 or $48,000 per year in savings respectively. The catch: Copilot is embedded in Word, Excel, and Teams in a way Quick isn’t. If your organization lives in Microsoft 365, Copilot’s integration depth is hard to match. If you’re multi-vendor — Slack, Google Workspace, Salesforce, and AWS — Quick’s 102 integrations https://aws.amazon.com/quick/pricing/ cover the landscape more broadly. The AWS Agent Stack Is Now Complete Quick is the third leg of a platform AWS has been assembling through 2026: Kiro is the agentic IDE for developers, Bedrock AgentCore is the infrastructure for deploying production agents, and Quick is the AI assistant for every other worker in the organization. Together they’re AWS’s answer to Microsoft’s parallel stack GitHub Copilot, Microsoft 365 Copilot, Azure AI and Google’s Gemini, Workspace AI, Vertex . The move to “no AWS account required” for Quick Free is the tell. AWS is trying to get into organizations from the user side — get individuals using Quick for free, then convert to enterprise contracts — rather than requiring IT to provision access first. It’s a consumer-style growth motion that’s new for AWS. Amazon Quick is available now https://aws.amazon.com/quick/ .