AWS Quick and Nous Research’s Hermes Agent both released updates focused on standardizing individual AI workflows into team-wide, reusable skills or flows. The changes target enterprise teams looking to move beyond ad-hoc agent usage toward governed, repeatable processes.
AWS Quick adds multi-agent orchestration and chat-built flows
AWS Quick introduced multi-agent orchestration, allowing a primary agent to delegate subtasks to several specialized agents simultaneously. The system merges their outputs into a single response while preserving each agent’s instructions, knowledge, and context. This architecture lets teams maintain separate, domain-specific agents while presenting unified results to the end user.
The platform also expanded its flow-building capabilities. Users can now construct, test, and refine automated business processes entirely through chat. The update adds deterministic steps to these flows, enabling fixed calculations and threshold checks that keep results consistent across runs. These flows support longer execution times and larger data volumes, moving beyond simple prompt-response tasks into structured data processing. A visual diagram displays each step alongside the chat interface, allowing users to verify the workflow matches their intended logic before publishing.
The updated Quick web app includes a daily briefing feed that prioritizes emails and messages requiring attention. Agents can now schedule tasks or initiate long-running requests before stepping away, continuing execution after the browser tab is closed. A new Library view consolidates recent work, pinned items, and generated files, syncing across web, desktop, and mobile clients.
Connectors for Meta ads and Amplitude now use the Model Context Protocol (MCP) server. These integrations allow teams to query ad performance and product usage data in natural language, combining third-party metrics with internal sources. The MCP server architecture standardizes how Quick communicates with external tools, allowing the platform to pull behavioral insights and campaign data into existing dashboards and workflows. The connectors enforce existing user permissions and require explicit approval before executing write actions, such as modifying ad budgets or creating campaigns.
Illustrative image: axonite / Pixabay
Administrative controls for prompt guardrails, group-level permissions, and data source restrictions are available to Professional and Enterprise subscribers. Administrators can filter content, block specific topics, and detect prompt attacks in detect-only or enforced modes. Custom permission profiles can be assigned to identity provider groups, and organizations can restrict which data sources users are allowed to create or share. The platform now supports Japanese, German, Chinese (Simplified and Traditional), and Vietnamese.
Real customers report automating complex business processes to make them more consistent and repeatable. One Latin American fintech provider noted that the platform allowed nearly four times as many people to work with data each month without growing the data team, saving 80 hours of work per month on a single use case.
Hermes Agent brings reusable skills to Microsoft Teams
Nous Research released an update for its open-source Hermes Agent, adding a native Microsoft Teams integration and a desktop application available through the Microsoft Store. The update introduces a feature called "Collective Wisdom," which automatically converts successful individual workflows into reusable skills composed of a description and code. After a review step, these skills can be shared with the team, standardizing how specific tasks are handled across the organization.
Hermes agents execute within Azure Container Apps Sandboxes, which function as isolated microVMs. This architecture ensures that collected knowledge and workflow data remain strictly within the company's environment and do not feed into any external AI provider's training data. The sandboxed execution model provides hardware-level isolation for agent processes, which is particularly relevant for organizations handling sensitive financial or legal documents. Nous Research, a portfolio company of Microsoft's M12 venture fund, reports that Hermes is the most widely used open agent on OpenRouter, processing nearly two trillion tokens daily.
The desktop application allows Windows users to run agents locally while maintaining the secure sandbox execution model. The Teams integration enables employees to hand off tasks such as financial analysis or inbox sorting directly within the chat interface, reviewing and adjusting results without switching applications.
Sources
- aws.amazon.com: New in AWS Quick: More capable flows, an updated web experience, and agents that work together | Amazon Web Se
- the-decoder.com: Hermes Agent's Collective Wisdom feature turns one employee's AI trick into the whole team's skill
PNEUMETRON EDITORIAL
Pneumetron's pipeline pairs AI-assisted drafting with human editorial review before publishing — our goal is to make staying informed easier, not to replace real reporting.




