Have a project in mind? Let's talk.

AWS Quick, Amazon's AI workspace for business work, received a substantial update on November 20 that pushes it further from a chat assistant toward an agent operations platform. The headline items: flows you can build and refine entirely in conversation, now with deterministic steps and longer runs; multi-agent orchestration that lets one request fan out to several specialized agents at once; a reworked web app that keeps agents working after you close the tab; new MCP-based connectors for Meta ads and Amplitude; and a preview of prompt guardrails for administrators. Source: New in AWS Quick, AWS Machine Learning Blog
AWS positions Quick as "the AI workspace built only for work," with chat-to-deliverable document and dashboard generation, delegated research, no-code internal apps, and connectors into tools like Slack, Salesforce, and Jira. This release is less about any single feature and more about a direction: individual prompting becoming team process. Source: Source
Quick Flows, which automate and standardize business processes, can now be built, run, and refined entirely in chat. You describe the process the way you would explain it to a new employee, Quick builds the flow, and a diagram beside the chat shows each step so you can confirm it matches how you work. Underneath, a flow orchestrates agents, skills, tools, and data, and the new deterministic steps, fixed calculations and checks, keep results consistent and verifiable across runs, with support for longer-running workflows and larger data volumes. Source: New in AWS Quick, AWS Machine Learning Blog
AWS's own walkthrough is deliberately ordinary: a finance manager turns a monthly variance report that lived in her memory and a personal checklist into a published flow her teammates can run, review, and approve from chat, with the flow pulling dashboard numbers, comparing regions to plan with set formulas, flagging outliers against thresholds she defined, and drafting commentary and an Excel summary for human review. The example is fictional, and AWS labels it as such; the customer quote it pairs with, from Latin American fintech Pomelo, claims nearly four times as many people now work with data each month and a single use case saving 80 hours of work per month. Those figures are the customer's as relayed by AWS, worth reading as directional rather than audited. Source: New in AWS Quick, AWS Machine Learning Blog
The updated Quick web app reorganizes around attention and continuity. A home-screen feed and daily briefing pull together the email and messages that need you in priority order, with options to ask Quick about an item or have it draft replies for review. A reorganized left navigation brings agents, flows, skills, knowledge, and connectors under one Customize section, and your conversations, agents, and files sync across web, desktop, and mobile. Source: New in AWS Quick, AWS Machine Learning Blog
The operationally significant part is asynchronous work: schedule a task or start a longer request, and Quick keeps working after you close the browser. When you return, results and any agents waiting on you surface in a view AWS calls Mission Control. For teams tracking the broader pattern, this is the same "agents keep running while you are away" model appearing across agent platforms this quarter, now landed in AWS's business workspace. A Library section gives agents and users a shared place for working files, rounding out the continuity story. Source: New in AWS Quick, AWS Machine Learning Blog
With multi-agent orchestration, the agent you ask can bring in several specialized agents at the same time, each working on the part it knows best, then combine their work into one response that shows what each agent contributed. AWS's example: asking why shipping costs rose in a region can pull in a forecasting agent built on planning data and a procurement agent built on vendor contracts. Each agent keeps its own instructions, knowledge, and context, so the people closest to that work build and maintain their own agent. Source: New in AWS Quick, AWS Machine Learning Blog
The design choice worth noting is the ownership model: specialists stay separately maintained rather than merged into one mega-agent. That mirrors how mature AgentOps teams already structure agent fleets, keeps the blast radius of a bad instruction small, and makes contribution visible in the response instead of hiding which agent said what. Source: New in AWS Quick, AWS Machine Learning Blog
Two new connectors arrive, both over the Model Context Protocol. The Meta ads connector, through Meta's ads MCP server, lets users create, manage, and analyze campaigns in natural language, with Quick asking for permission before it creates a campaign, changes a budget, or takes any other write action in an ad account. The Amplitude connector, also through its MCP server, answers product usage, behavior, and conversion questions in natural language. Across connectors, Quick respects existing permissions, shows each person only what they already have access to, and administrators choose which connectors teams can add. Source: New in AWS Quick, AWS Machine Learning Blog
The administrative additions are as notable as the user-facing ones. Guardrails for prompts, in preview, let administrators filter content, block denied topics and words, detect prompt attacks, and protect selected types of personal information in the prompts people send to Quick, with a detect-only mode before enforcement and scoping to individual users, identity provider groups, or the whole account. Permissions and limit profiles can now be assigned by identity provider group, so an admin can prevent a finance group from creating agents or cap how many agent hours each member of a group can use. Administrators can also control which data source types people can create, update, or share, for example allowing Amazon Redshift while blocking file uploads. All three are available on Professional and Enterprise plans. Source: New in AWS Quick, AWS Machine Learning Blog
The release rounds out with skills in the web app, reusable instructions that teach Quick to do a task your way, such as reviewing a contract against a team checklist or writing an update in the company format, and new language support for Japanese, German, Chinese in Simplified and Traditional variants, and Vietnamese. All features are available now in every Region where AWS Quick is offered. Source: New in AWS Quick, AWS Machine Learning Blog
Our assessment is that the chat-built flows will get the demo attention, but the governance trio, prompt guardrails with detect-only mode, group-scoped permissions with agent-hour caps, and data-source controls, is what makes the rest deployable in a regulated environment. Deterministic steps inside agentic flows answer the "same result every run" objection that has kept many finance and operations teams away from agent automation. This connects to themes we have covered: the managed tool approvals pattern is the same permission-before-write idea Quick applies to ad accounts, and the AWS Loom agent platform vulnerabilities post is a reminder that governance features and their enforcement gaps both deserve review before rollout.
AWS Quick is Amazon's AI workspace for business work: it turns conversations into documents, spreadsheets, decks, and dashboards, delegates research, builds no-code internal apps, and connects to tools like Slack, Salesforce, and Jira through connectors that respect existing permissions. Source: AWS Quick product page
Five headline items: flows built and refined in chat with deterministic steps and longer runs, multi-agent orchestration across specialized agents, a reworked web app with an attention feed and asynchronous work, Meta ads and Amplitude connectors over MCP, and a preview of prompt guardrails for administrators. Source: New in AWS Quick, AWS Machine Learning Blog
One agent can bring in several specialized agents simultaneously, each handling the part it knows best with its own instructions, knowledge, and context. Quick combines their work into a single response that shows what each agent contributed. Source: New in AWS Quick, AWS Machine Learning Blog
For the Meta ads connector, Quick asks for permission before creating a campaign, changing a budget, or any other write action in an ad account. Generally, connectors show each user only what they already have access to, and administrators choose which connectors teams can add. Source: New in AWS Quick, AWS Machine Learning Blog
Prompt guardrails in preview with a detect-only mode, permissions and agent-hour limits assignable by identity provider group, and controls over which data source types people can create, update, or share, all on Professional and Enterprise plans. Source: New in AWS Quick, AWS Machine Learning Blog
Ready to take the first step towards unlocking opportunities, realizing goals, and embracing innovation? We're here and eager to connect.