AIToday
Large Language ModelsAI Business & IndustryAmazon AI BlogPublished: Oct 15, 2026, 01:00 JST

AWS: Amazon Bedrock AgentCore builds decks from docs under 90 seconds

AWS: Amazon Bedrock AgentCore builds decks from docs under 90 seconds

3 Key Points

  1. What happened

    AWS published a guide and code for deploying a supervisor agent on Amazon Bedrock AgentCore runtime that turns enterprise documents into PowerPoint presentations, completing in under 90 seconds in internal testing. The code is in a public repository.

  2. Why it matters

    A financial services team in the guide spends significant time each week manually building client PowerPoint presentations from quarterly reports, compliance documents, and market analyses. This solution completes the task in under 90 seconds, which could reduce that manual work.

WHO IT HITSThis lands on enterprise IT teams and developers at financial services, engineering, and sales organizations who deploy AI agents on AWS. They would need to implement the solution themselves using the public repository, since no finished product is being sold.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

The guide's example workload is a financial services team that spends significant time each week manually building client PowerPoint presentations from quarterly reports, compliance documents, and market analyses. The solution automates that workflow using a supervisor agent on Amazon Bedrock AgentCore runtime, with the code in a public repository.

In internal testing with a 90-page Amazon 2024 Annual Report PDF, the solution produced a six-slide presentation in 70–90 seconds end-to-end. That includes knowledge base retrieval, content generation, image creation, and file assembly. Image generation used Stability AI Stable Image Inpaint, and the guide says it accounted for the majority of the approximately $0.04 in model invocation costs per generation at time of writing in us-east-1.

The solution addresses a technical bottleneck: the default Amazon API Gateway integration timeout is 29 seconds, but presentation generation takes 70–90 seconds. The guide instructs users to configure the timeout to at least 300 seconds. It also notes that current manual document preparation involves reading hundreds of pages and extracting key data points, structuring slides, and generating visuals.

FAQ
How long does the presentation generation take?
In internal testing with a 90-page Amazon 2024 Annual Report PDF, the solution produced a presentation in 70–90 seconds end-to-end. The editor's summary says it completes in under 90 seconds.
What does it cost to generate a presentation?
The guide says cost per generation is approximately $0.04 in model invocation costs at time of writing in us-east-1. Image generation accounts for the majority of that cost.
Can I use my own model?
Yes. The configuration file lets you adjust model identifiers for parsing, agent reasoning, embedding, and image generation. The guide recommends Anthropic Claude Haiku 4.5 for cost efficiency in parsing and Anthropic Claude Sonnet 4.6 for agent reasoning.
Amazon AI BlogRead Original Article

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleThird Moment runs 10,000 agents for 36-48 hours using GPT-6.1 Sol