According to an announcement by Hillary Lovric, Senior Product Marketing Manager for AWS Quick, the company has launched updates to its workspace built only for work. The updates include the ability to build and improve business processes through chat conversation, an updated web app, multi-agent orchestration, new data source connectors, and security and governance controls.
Building, Running, and Refining Business Processes in Conversation
According to AWS's publication, AWS Quick Flows enables teams to build, run, and refine workflows (Flows) directly from the chat interface across the web, desktop, or mobile applications. Users describe a process in their own words, and the system builds a diagram of the workflow steps displayed alongside the conversation for testing and confirmation. Workflows now include deterministic steps designed to ensure consistent and verifiable calculations and checks, as well as support for longer-running processes and larger data volumes.
The announcement illustrates these capabilities through an example scenario involving an FP&A (financial planning and analysis) manager named Maya at a retailer. Rather than using a manual process where monthly results are pulled from a dashboard and compared manually against the budget, the flow configured in Quick pulls data from the monthly dashboard, processes data across multiple geographic regions in parallel, compares them to the annual plan in the budget spreadsheet using fixed formulas, and flags discrepancies based on predefined thresholds. Subsequently, an agent drafts explanatory commentary using a dedicated commentary skill, generates a summary draft in Microsoft Excel for the shared space and email, and alerts the team in Slack. Maya publishes the flow for her team. At the next month-end, Jordan, an analyst on the team, runs it from chat in the Quick desktop app, where the flow can use a spreadsheet of regional adjustments saved on his computer. Before the report is sent, Jordan reviews the draft, including why each region was flagged, requests a few changes in chat, and approves it for the VP of Finance. When Priya joins the team, she runs the same flow at her first month-end, and the report is built the same way.
Rodrigo Marini, Head of Data at Pomelo—a Latin American fintech provider powering card issuance, processing, and payment operations—stated that using AWS Quick changed who gets to build solutions within the organization. According to Marini, previously every analysis and automation depended on the data team, whereas today employees across all departments build agents and flows within a unified and governed environment. Marini noted that nearly four times as many employees work with data each month without expanding the team, a single use case saves approximately 80 hours of work per month, and the data team can focus on its most complex tasks.
Web App Upgrades and Ongoing Task Management
The updated Quick web app consolidates items that need user attention on the home screen alongside a daily briefing. The left navigation bar was reorganized under the Customize category, bringing together agents, flows, skills, knowledge bases, and connectors, including components built on the desktop version or shared by other users. Existing dashboards, topics, and spaces continue to function as before.
The system allows users to schedule tasks or initiate long-running requests before meetings, with agents continuing to execute even after the browser tab is closed. Upon returning, the user can review results and respond to agents waiting for them in Mission Control. The Library area consolidates recent items, favorites, pinned items, and documents created within the system, while synchronizing conversations, agents, and files across the browser, desktop, and mobile devices.
Multi-Agent Orchestration and New Connectors for Data Analysis
The update introduces multi-agent orchestration. Through a single prompt, a central agent can trigger multiple specialized agents in parallel, with each agent operating according to its defined instructions, knowledge, and context. The system merges their findings into a single response detailing what each agent contributed. For example, an inquiry regarding an increase in shipping costs in a specific region can simultaneously invoke a forecasting agent based on planning data and a procurement agent based on vendor contracts.
In addition, AWS added new connectors that support natural language queries:
- Meta ads: A connection to the Meta ads Model Context Protocol (MCP) server, enabling users to create, manage, and analyze campaigns, and integrate this information with additional data sources. The system requests explicit user permission before executing any write action, such as creating a campaign or altering a budget.
- Amplitude: A connection to the Amplitude MCP server to pull behavioral insights, usage data, and customer conversions directly into dashboards, documents, and flows.
All connectors operate in accordance with existing user permissions, write actions require prior authorization, and system administrators determine which connectors can be added across the organization.
Skills on Web, Security Controls, and Additional Languages
The new release includes support for skills directly within the web app, allowing users to create and manage reusable instructions, such as reviewing a contract against a checklist or drafting a project update.
For customers subscribed to Professional and Enterprise plans, new security and governance controls have been added:
- Guardrails for prompts (preview): Administrators can set policies to filter content, block denied words and topics, detect prompt attacks, and protect selected types of personal information. Policies can be tested in detect-only mode prior to enforcement and applied to individual users, identity provider (IdP) groups, or the entire account.
- Permissions and limits by group: Configuration of custom permissions and usage limits (such as agent hours) for identity provider groups without requiring individual user management.
- Data source controls: Defining the types of data sources users can create, update, or share, such as permitting Amazon Redshift connections while blocking file uploads.
The system now supports additional languages: Japanese, German, Chinese (Simplified and Traditional), and Vietnamese. According to the announcement, customer data remains within their environment and is not used to train third-party models. These capabilities are available across all regions where AWS Quick is currently supported.