AWS cut marketing page production from hours to minutes with agentic AI

AWS shared how their marketing team converted page content assembly into a semi-automatic workflow using agentic AI. The solution on Amazon Bedrock, developed with Gradial, reduced the preparation time for a single web page from approximately four hours to ten minutes without sacrificing quality checks.

Where the process slowed down

AWS describes a typical scenario for a marketer this way: campaign brief, call with the digital team, task added to the backlog, then manual page assembly in the CMS. Even when the task itself is straightforward, time is consumed by block configuration, content placement, approvals, and revisions. The problem is especially noticeable in a corporate environment, where a single page must simultaneously comply with brand guidelines, SEO requirements, accessibility, and internal publication procedures for release on global digital platforms.

• Manual assembly of blocks and layouts in the CMS
• Delays caused by text, links, and creative reviews
• Dependence on engineers when ready-made components are insufficient
• Late-stage SEO, accessibility, and brand standard checks

AWS's key insight is that this isn't a collection of separate minor issues, but one systemic process failure. Quality was controlled too late, and the bulk of time went not to marketing strategy but to mechanical page assembly. As a result, marketers spent hours on operations that could be formalized: select the needed components, assemble the structure, run checks, and hand off the page for publication without a long chain of manual actions.

How the solution works

The new system is built around Amazon Bedrock and Anthropic Claude and Amazon Nova models. A marketer describes the task in natural language: which page needs to be assembled, which blocks are needed, and what the output should be. Next, the Gradial agent interprets the request, determines the page structure, selects components, and generates a configuration that previously required knowledge of the CMS's internal logic. Essentially, the interface for the author transforms from a set of complex forms into a dialogue, where the system itself breaks down the task into steps.

A separate layer in the architecture is an MCP server for real-time quality checks. Instead of waiting for final review, the system compares content with SEO, accessibility, and brand standard requirements as the page is being assembled. If an image, text, or page structure fails validation, the problem is visible immediately in the same session. This eliminates the characteristic corporate marketing loop where a page is almost ready but then rolls back for revision due to a single incorrect element.

The final stage is programmatic submission of the result to the corporate CMS through a proxy layer. It doesn't replace the publishing system, but connects the agent to the existing infrastructure so the page is created within the familiar data model and governance rules. Because of this, AWS didn't have to rebuild the entire publishing process from scratch: the agent automates assembly and handoff, while the control environment of the CMS remains in place with the necessary permissions, logging, and approval stages.

What changed after

After launch in production, AWS compared metrics before and after implementation. The most noticeable metric is page assembly time: instead of manual work lasting up to four hours, it was reduced to approximately ten minutes—a reduction of more than 95%. But the time savings are not the only benefit. Quality checks became proactive, the interface for the team became simpler, and the process itself became more linear. Where there were once separate stages of configuration, review, and handoff, most actions now take place in a single stream. For marketing teams, this means time is freed up for tasks that truly affect campaign results: positioning, messaging, hypothesis testing, and content optimization.

AWS directly formulates the project's goal this way: eliminate mechanical work that doesn't create value on its own. If the approach scales to other digital channels, agentic AI could become not just an accelerator for CMS but a new operational layer for content teams in an enterprise environment.

What this means

The AWS case demonstrates a more grounded scenario for agentic AI than the typical chatbot demonstrations: not generation for its own sake, but automation of a specific bottleneck in a business process. If an agent can understand the task, assemble a page, validate the result, and work with the existing CMS, the company gets not an experiment but a measurable tool with clear ROI and direct connection to team operational metrics, launch speed, and publication cost.