Change
fe4b438
fe4b438d073de76df847392ebf72383bbfbcbdd2 · commit on GitHub
aws-blog-feed: changed (548236 bytes, HTTP 200)
raw/aws-blog-feed/response.xml modified
- Source
- aws-blog-feed
- Lines added
- +291
- Lines removed
- -198
- Stored bytes at this commit
- 548,236
- Timestamp
- observed
- Raw artifact at this commit
- raw/aws-blog-feed/response.xml
Recorded headers
| observed_at | 2026-10-16T04:32:13.757Z |
|---|---|
| origin_date | null |
| status | 200 |
| final URL | https://aws.amazon.com/blogs/machine-learning/feed/ |
| etag | null |
| last-modified | Fri, 16 Oct 2026 02:46:33 GMT |
| date | Fri, 16 Oct 2026 04:32:13 GMT |
| age | null |
| cache-control | null |
| cf-cache-status | null |
| content-encoding | null |
| content-length | null |
@
@@ -5,7 +5,7 @@ <atom:link href="https://aws.amazon.com/blogs/machine-learning/feed/" rel="self" type="application/rss+xml"/> <link>https://aws.amazon.com/blogs/machine-learning/</link> <description>Official Machine Learning Blog of Amazon Web Services</description>-
<lastBuildDate>Wed, 14 Oct 2026 17:10:11 +0000</lastBuildDate>+
<lastBuildDate>Thu, 15 Oct 2026 17:43:35 +0000</lastBuildDate> <language>en-US</language> <sy:updatePeriod> hourly </sy:updatePeriod>@
@@ -13,6 +13,295 @@ 1 </sy:updateFrequency> <item>+
<title>Meta’s Muse Spark Open now available in Amazon SageMaker JumpStart</title>+
<link>https://aws.amazon.com/blogs/machine-learning/metas-muse-spark-open-now-available-in-amazon-sagemaker-jumpstart/</link>+
+
<dc:creator><![CDATA[Aris Tsakpinis]]></dc:creator>+
<pubDate>Thu, 15 Oct 2026 17:43:35 +0000</pubDate>+
<category><![CDATA[Amazon SageMaker AI]]></category>+
<category><![CDATA[Amazon SageMaker JumpStart]]></category>+
<category><![CDATA[Announcements]]></category>+
<category><![CDATA[Intermediate (200)]]></category>+
<guid isPermaLink="false">8370f01f72d733de10ec1ce843e1456b54b19515</guid>+
+
<description>Muse Spark Open, Meta's coding-optimized open-weight model, is now available in Amazon SageMaker JumpStart. Deploy it to a SageMaker AI endpoint in your own AWS account using SageMaker Studio or the SageMaker Python SDK, and run your first inference request.</description>+
<content:encoded><p>Muse Spark Open, Meta’s coding-optimized open-weight model, is now available in <a href="https://aws.amazon.com/sagemaker/ai/jumpstart/" target="_blank" rel="noopener">Amazon SageMaker JumpStart</a>, the machine learning (ML) hub of <a href="https:/…+
<p>In this post, we introduce Muse Spark Open and show you how to deploy the model with <a href="https://aws.amazon.com/sagemaker/ai/studio/" target="_blank" rel="noopener">Amazon SageMaker Studio</a> and the <a href="https://sagemaker.readthedocs.io/en/stable/" target="_blank" …+
<h2 id="overview-of-muse-spark-open">Overview of Muse Spark Open</h2>+
<p>Meta describes Muse Spark Open as a coding-optimized model trained for agentic coding workflows, with a focus on first-attempt accuracy, tool calling, and understanding of large codebases. It’s a reasoning model, and it accepts text, images, documents, audio, and audio-video as input and re…+
<p>Muse Spark Open uses a mixture-of-experts (MoE) architecture with 816 billion total parameters, of which 42 billion are active for each token. The model holds the capacity of a very large network, while the compute it spends on each token is closer to that of a much smaller dense model. Met…+
<p>The following table summarizes the model.</p>+
<table border="1px" width="100%" cellpadding="10px">+
<tbody>+
<tr>+
<td><strong>Attribute</strong></td>+
<td><strong>Muse Spark Open</strong></td>+
</tr>+
<tr>+
<td>Provider</td>+
<td>Meta</td>+
</tr>+
<tr>+
<td>Architecture</td>+
<td>Mixture of experts (MoE), reasoning model</td>+
</tr>+
<tr>+
<td>Parameters</td>+
<td>816B total, 42B active</td>+
</tr>+
<tr>+
<td>Context length</td>+
<td>1M tokens</td>+
</tr>+
<tr>+
<td>Input modalities</td>+
<td>Text, image, document, audio, audio-video</td>+
</tr>+
<tr>+
<td>Output modalities</td>+
<td>Text</td>+
</tr>+
<tr>+
<td>Variants in SageMaker JumpStart</td>+
<td>BF16 and NVFP4</td>+
</tr>+
<tr>+
<td>License</td>+
<td>Muse Model License, including the Muse Spark Usage Policy</td>+
</tr>+
</tbody>+
</table>+
<h2 id="built-for-long-running-coding-agents">Built for long-running coding agents</h2>+
<p>Meta trained Muse Spark Open on the way developers work in real repositories: code generation, debugging, multi-file refactors, and tasks that run for a long time. With the 1 million-token context window, the model can take in a whole project, including its dependencies, legacy code, and th…+
<p>Meta also designed Muse Spark Open as an agentic model from the start. Meta says the model plans, acts, observes the result, and corrects itself over long horizons, rather than making single tool calls. As a reasoning model, it works through a task before it acts, and the SageMaker AI endpo…+
<p>Two properties of the model shape how you serve it. First, sparse activation keeps per-token compute low, but serving still requires enough GPU memory to hold all 816 billion parameters, so the instance choice comes down to memory capacity. Second, the NVFP4 variant stores weights in NVIDIA…+
<p>With open weights on SageMaker AI, you also control the deployment. The endpoint runs in your AWS account, so your source code and prompts are processed on infrastructure that you configure and control. You can attach the endpoint to your own <a href="https://aws.amazon.com/vpc/" target=…+
<h2 id="use-cases">Use cases</h2>+
<p>You can use Muse Spark Open for a wide range of applications. Meta’s focus on coding and agentic workflows suits it to the following use cases:</p>+
<ul>+
<li><strong>Coding assistants over private repositories</strong>: Answer questions about large codebases, refactor them, and debug them with the full project in context, while the code is processed on an endpoint in your account.</li>+
<li><strong>Long-running coding agents</strong>: Run agents that plan a change, edit across multiple files, run tests in your own sandbox, and correct course based on the results.</li>+
<li><strong>Multimodal engineering workflows</strong>: Turn screenshots, design documents, or a recorded bug report with audio and video into a reproduction plan or a draft fix.</li>+
</ul>+
<h2 id="getting-started-with-sagemaker-jumpstart">Getting started with SageMaker JumpStart</h2>+
<p>You can deploy Muse Spark Open from SageMaker Studio or with the SageMaker Python SDK. Each path creates a real-time SageMaker AI endpoint in your account.</p>+
<h3 id="prerequisites">Prerequisites</h3>+
<p>Before you start, make sure that you have the following:</p>+
<ul>+
<li>An AWS account with permissions to create SageMaker AI models and endpoints.</li>+
<li>An <a href="https://aws.amazon.com/iam/" target="_blank" rel="noopener">AWS Identity and Access Management (IAM)</a> execution role for SageMaker AI. For this walkthrough, the <code>AmazonSageMakerFullAccess</code> managed policy is sufficient. For production, scop…+
<li>A SageMaker AI domain with access to SageMaker Studio, for the Studio path. For setup instructions, refer to <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/onboard-quick-start.html" target="_blank" rel="noopener">Use quick setup for Amazon SageMaker AI</a>.</li>…+
<li>Service quota for the endpoint instance type: <strong>ml.p5e.48xlarge for endpoint usage</strong> for the NVFP4 variant, or <strong>ml.p6-b300.48xlarge for endpoint usage</strong> for the BF16 variant. To raise a quota, refer to <a href="https://docs.aws.amazon.…+
<li>For the SDK path: Python 3.10 or later and SageMaker Python SDK v3 (<code>pip install --upgrade sagemaker</code>).</li>+
<li>An AWS Region where Muse Spark Open is available in SageMaker JumpStart. At launch, Muse Spark Open is available in US East (N. Virginia) and US East (Ohio). To check availability in other Regions, open SageMaker JumpStart in your Region.</li>+
</ul>+
<p>The instances that host Muse Spark Open are multi-GPU instances, and you pay for an endpoint for as long as it runs, whether or not it serves requests. Delete the endpoint when you finish, as described in the clean-up section of this post. For rates, refer to <a href="https://aws.amazon.…+
<h3 id="deploy-using-sagemaker-studio">Deploy using SageMaker Studio</h3>+
<p>The SageMaker Studio path suits a first evaluation: you find the model, review its license, and deploy it from one model detail card.</p>+
<p><strong>To find Muse Spark Open in SageMaker JumpStart</strong></p>+
<ol type="1">+
<li>Open SageMaker Studio.</li>+
<li>In the navigation pane, choose <strong>JumpStart</strong>.</li>+
<li>In the search box, enter Muse Spark Open.</li>+
<li>Choose <strong>Muse-Spark-Open-816B-A42B</strong> for the BF16 variant or <strong>Muse-Spark-Open-816B-A42B-NVFP4</strong> for the NVFP4 variant to open its model detail card.</li>+
</ol>+
<p>Figure 1 shows the model detail card, which lists the model description, license, and the <strong>Deploy</strong> option.</p>+
<div style="width: 810px" class="wp-caption alignnone">+
<a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-1.jpg" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-1.jpg" alt="Muse Spark Open model d…+
<p class="wp-caption-text">Figure 1: The Muse Spark Open model detail card in SageMaker JumpStart</p>+
</div>+
<p><strong>To deploy the model</strong></p>+
<ol type="1">+
<li>On the model detail card, choose <strong>Deploy</strong>.</li>+
<li>Muse Spark Open is a gated model, so SageMaker Studio shows its end-user license agreement (EULA). Review the agreement and accept it to continue.</li>+
<li>For <strong>Endpoint name</strong>, keep the generated name or enter your own.</li>+
<li>For <strong>Instance type</strong>, select <strong>ml.p5e.48xlarge</strong> for NVFP4 or <strong>ml.p6-b300.48xlarge</strong> for BF16. SageMaker Studio lists only the instance types that the model supports.</li>+
<li>For <strong>Initial instance count</strong>, enter 1.</li>+
<li>(Optional) Under <strong>Advanced options</strong>, select a VPC and an AWS KMS key, or turn on network isolation.</li>+
<li>Choose <strong>Deploy</strong>.</li>+
</ol>+
<p>Figure 2 shows the deployment settings for the NVFP4 variant.</p>+
<div style="width: 810px" class="wp-caption alignnone">+
<a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-2.jpg" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-2.jpg" alt="Deployment settings for…+
<p class="wp-caption-text">Figure 2: Deployment settings for Muse Spark Open</p>+
</div>+
<p>Deployment typically takes 15–20 minutes while SageMaker AI provisions the instance and loads the weights. When the endpoint status changes to <strong>InService</strong>, the endpoint is ready for requests, as shown in Figure 3.</p>+
<div style="width: 810px" class="wp-caption alignnone">+
<a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-3.png" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-3.png" alt="Muse Spark Open endpoin…+
<p class="wp-caption-text">Figure 3: The Muse Spark Open endpoint in service</p>+
</div>+
<h3 id="deploy-using-the-sagemaker-python-sdk">Deploy using the SageMaker Python SDK</h3>+
<p>SageMaker JumpStart identifies each variant by a model ID:</p>+
<ul>+
<li>BF16: <code>meta-vlm-muse-spark-open-816b-a42b</code>.</li>+
<li>NVFP4: <code>meta-vlm-muse-spark-open-816b-a42b-nvfp4</code>.</li>+
</ul>+
<p>The following code deploys the NVFP4 variant with SageMaker Python SDK v3. You can run it in a SageMaker Studio notebook, which uses your domain’s execution role. In any other Python environment with AWS credentials, pass your SageMaker AI execution role with the <code>role_arn</co…+
<div class="hide-language">+
<pre><code class="language-python">from sagemaker.core.jumpstart.configs import JumpStartConfig+
from sagemaker.serve import ModelBuilder+
from sagemaker.train.configs import Compute+
+
# NVFP4 variant; for BF16, use "meta-vlm-muse-spark-open-816b-a42b" with ml.p6-b300.48xlarge+
model_id = "meta-vlm-muse-spark-open-816b-a42b-nvfp4"+
+
jumpstart_config = JumpStartConfig(+
model_id=model_id,+
accept_eula=True, # Confirms that you accept Meta's license terms+
)+
+
model_builder = ModelBuilder.from_jumpstart_config(+
jumpstart_config=jumpstart_config,+
compute=Compute(instance_type="ml.p5e.48xlarge"),+
# role_arn="&lt;your-sagemaker-execution-role-arn&gt;", # Required outside SageMaker Studio+
)+
+
model = model_builder.build(model_name="muse-spark-open-nvfp4")+
endpoint = model_builder.deploy(endpoint_name="muse-spark-open-nvfp4")</code></pre>+
</div>+
<h3 id="run-inference">Run inference</h3>+
<p>When the endpoint is in service, send it a request with the <code>invoke</code> method. The endpoint accepts an OpenAI-compatible chat payload with a <code>messages</code> list and returns a chat completion. The final answer is in <code>message.content</code…+
<div class="hide-language">+
<pre><code class="language-python">import json+
+
payload = {+
"messages": [+
{+
"role": "user",+
"content": "Write a Python function that validates ISO 8601 dates, and add pytest unit tests for it.",+
}+
],+
"max_tokens": 4096,+
}+
+
response = endpoint.invoke(+
body=json.dumps(payload),+
content_type="application/json",+
)+
+
result = json.loads(response.body.read().decode("utf-8"))+
message = result["choices"][0]["message"]+
print("Reasoning:", message.get("reasoning_content"))+
print("Answer:", message["content"])</code></pre>+
</div>+
<p>The endpoint also supports streaming. To receive tokens as they’re generated, add <code>"stream": true</code> to the request body and call <code>endpoint.invoke_with_response_stream</code> instead of <code>invoke</code>.</p>+
<h2 id="clean-up">Clean up</h2>+
<p>To stop incurring charges, delete the endpoint when you no longer need it. If you deployed with the SDK, delete the model, the endpoint, and the endpoint configuration:</p>+
<div class="hide-language">+
<pre><code class="language-python">from sagemaker.core.resources import EndpointConfig+
+
endpoint_config = EndpointConfig.get(endpoint_config_name=endpoint.endpoint_name)+
model.delete()+
endpoint.delete()+
endpoint_config.delete()</code></pre>+
</div>+
<p>If you deployed from SageMaker Studio, delete the endpoint in the SageMaker AI console: choose <strong>Inference</strong>, then <strong>Endpoints</strong>, select the endpoint, and choose <strong>Actions</strong>, <strong>Delete</strong>.</…+
<h2 id="conclusion">Conclusion</h2>+
<p>In this post, we showed how to deploy Muse Spark Open, Meta’s coding-optimized open-weight model, from SageMaker JumpStart to an endpoint in your own AWS account. You used SageMaker Studio or the SageMaker Python SDK and sent the model a first inference request. Benchmark both variants on y…+
<p>To get started, search for Muse Spark Open in SageMaker JumpStart in SageMaker Studio. For more deployment options, such as inference components and optimized deployment configurations, refer to <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/jumpstart-foundation-models-use.html…+
<p>&nbsp;</p>+
<p style="clear: both"></p>+
<hr style="width: 100%">+
<h2>About the authors</h2>+
<footer>+
<div class="blog-author-box" style="padding-top: 2.0em">+
<div class="blog-author-image" style="margin-right: 1.0em">+
<img loading="lazy" class="alignnone size-full" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-4.jpg" alt="Aris Tsakpinis" width="100" height="133">+
</div>+
<h3 class="lb-h4">Aris Tsakpinis</h3>+
<p style="overflow: hidden"><a href="https://www.linkedin.com/in/aristsakpinis/" target="_blank" rel="noopener">Aris</a> is a Senior Specialist Solutions Architect for Generative AI focusing on open-weight models on Amazon Bedrock and the broader generative AI open-source ecosyst…+
</div>+
<div class="blog-author-box" style="padding-top: 2.0em">+
<div class="blog-author-image" style="margin-right: 1.0em">+
<img loading="lazy" class="alignnone size-full" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/ML-22194-5.jpg" alt="Liam Neal Reilly" width="100" height="133">+
</div>+
<h3 class="lb-h4">Liam Neal Reilly</h3>+
<p style="overflow: hidden"><a href="https://www.linkedin.com/in/liam-neal-reilly/" target="_blank" rel="noopener">Liam</a> is a Software Engineer on the Amazon SageMaker JumpStart team at AWS, where he focuses on deploying and optimizing open-weight AI models. He received his un…+
</div>+
</footer></content:encoded>+
+
+
+
</item>+
<item>+
<title>Trust in Depth: Giving agents more autonomy without losing control</title>+
<link>https://aws.amazon.com/blogs/machine-learning/trust-in-depth-giving-agents-more-autonomy-without-losing-control/</link>+
+
<dc:creator><![CDATA[Shawn Bice]]></dc:creator>+
<pubDate>Thu, 15 Oct 2026 15:05:35 +0000</pubDate>+
<category><![CDATA[Amazon Machine Learning]]></category>+
<category><![CDATA[Artificial Intelligence]]></category>+
<category><![CDATA[Intermediate (200)]]></category>+
<category><![CDATA[Thought Leadership]]></category>+
<guid isPermaLink="false">379390b4bb32581d87fbb3af3ac3000f67e1f7df</guid>+
+
<description>Enterprise leaders want AI agents to take on more work but don't fully trust them to act correctly every time. Trust in Depth is AWS's layered approach-scope, know, bound, observe, and prove-that lets agents act on their own when their actions can be verified, reserving human judgm…+
<content:encoded><p>In every conversation I have with enterprise leaders, they say the same thing. They want agents to do more, but they don’t trust them to do it right all the time.</p>+
<p>That’s fair. The same creativity and autonomy that makes agents useful makes them hard to trust. Give an agent a task and it finds its own path to finish it, picking tools and sequencing steps no developer scripted. Run the same task twice and it might take a different route each time. A te…+
<p>The usual answer is a person in the loop, approving everything the agent does. That works at small scale, but as agents take on more work, human approvals pile up and review fatigue sets in. People asked to check everything stop catching the things that matter. Caution becomes a bottleneck,…+
<p>The fix is to let the agent act on its own when its actions can be verified, and reserve human judgment for the decisions that need it. As agents take on higher stakes work, like financial transactions, a wrong action becomes increasingly harmful. Organizations need the confidence to hand a…+
<p>Content filtering, testing, and monitoring each serve an important purpose, but none of these capabilities alone can promise an agent will behave correctly the next time it runs. Trust comes from independent layers that reinforce one another, the same layered discipline the industry has pra…+
<h2 id="trust-in-depth-trust-you-can-verify"><strong>Trust in Depth: Trust you can verify</strong></h2>+
<p>You build trust by answering a set of questions at different moments in the agent’s lifecycle:</p>+
<ul>+
<li><strong>Scope, before it runs:</strong> What is this agent for, what capabilities does the job require, and what must it never do?</li>+
<li><strong>Know, at deployment:</strong> Who authorized it, what identity does it carry, and can you account for every agent acting on your behalf?</li>+
<li><strong>Bound, while it runs:</strong> Do its actions stay within the limits you set, and can it still do the job?</li>+
<li><strong>Observe, after each run:</strong> Can you reconstruct what the agent did and why, and learn from it?</li>+
</ul>+
<p>Across these questions, some requirements can be expressed as precise rules, while others call for judgment. A company might require that only a manager can approve access, or that an employee’s sign-in must be blocked before their mailbox can be shared. Other expectations, such as keeping …+
<p>For precisely defined rules, testing shows whether the controls work in the cases you anticipate, but an agent takes paths you didn’t. The technology that goes further is automated reasoning, the mathematical logic that checks whether a rule holds in every possible case. It can establish th…+
<p>AWS has <a href="https://www.amazon.science/blog/a-decade-of-mathematical-certainty-reflections-on-the-automated-reasoning-group" target="_blank" rel="noopener">run automated reasoning in production for more than a decade</a>, from Identity and Access Management (IAM) at more th…+
<p>However, not every rule can be stated precisely. “Only a manager can approve access” can be proven. “Keep a friendly tone” cannot. Where proof does not apply, limit the agent’s access and actions, evaluate its behavior, and send genuine ambiguity to a person to apply judgment. Grant more au…+
<p>Here is what each Trust in Depth layer asks of you, and how AWS helps you answer it.</p>+
<p><img loading="lazy" class="alignnone size-full wp-image-141442" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/15/trust-in-depth-hero-image-v2.png" alt="" width="1946" height="1093"></p>+
<h3>Scope your agent to only what the job requires</h3>+
<p>Trust begins before an agent runs, and this is where to invest the most effort. Problems caught here are cheap to fix. Scoping means deciding what work the agent owns, which tools, data, and services it may reach, the rules it must follow, and which decisions stay with a person.</p>+
<p>“Resolve the customer’s issue” says what to achieve, not how the agent may act to achieve it. The agent still needs limits on what it may see, what it may do, and when it must stop and ask. Give it only the access its job requires, and no more. Watch for capabilities no one granted on purpo…+
<p>Scoping starts at the build stage, with how you define the agent. <a href="https://kiro.dev/" target="_blank" rel="noopener">Kiro</a>, the agentic coding system, brings rigor and mature engineering practices to the build stage of agents. It turns high-level intent into structure…+
<p>The next step is turning those boundaries into rules you can hold the agent to. <a href="https://aws.amazon.com/bedrock/agentcore/">Amazon Bedrock AgentCore</a> provides governance controls to scale agents safely and governs what the agent does. Using Policies in Bedrock AgentCo…+
<p>Scoping your agent is the hard part, because turning human intent into durable, enforceable policy requires precision and completeness. However, it is also the part that pays off most. Everything that builds on what you set here, and the more precise the policy, the more your agent can safe…+
<h3 id="know-every-agent-acting-on-your-behalf-and-under-whose-authority">Know every agent acting on your behalf, and under whose authority</h3>+
<p>You can’t govern what you can’t see. Teams are creating agents faster than anyone can track them, and an agent no one knows about is an agent no one controls. Every agent needs its own identity, checked every time it acts. No action should be orphaned. When something goes wrong, you can tel…+
<p><a href="https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/registry.html" target="_blank" rel="noopener">AWS Agent Registry</a> provides a managed, approved catalog of the agents, Model Context Protocol (MCP) servers, tools, and skills across your organization. Admin…+
<h3 id="bound-what-your-agent-says-and-does-within-the-limits-you-set">Bound what your agent says and does, within the limits you set</h3>+
<p>Setting boundaries matters because an agent can take 10 steps you approved and still reach an outcome no one intended. It can expose sensitive data, drift off topic, or combine permitted actions in ways your policies forbid. Limits can fail in both directions. Too loose, and the agent can d…+
<p>This is where the policy you defined in Scope takes effect. <a href="https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/policy.html" target="_blank" rel="noopener">Policy in AgentCore</a> enforces it on every tool call routed through <a href="https://aws.amazon.com…+
<p>Boundaries also govern what an agent can reach in the first place: which files it can touch, which services it can call, and which credentials it can use. <a href="https://strandsagents.com/blog/strands-box-the-big-picture/" target="_blank" rel="noopener">Strands Box</a>, an ope…+
<h3 id="observe-what-your-agent-did-then-improve-it">Observe what your agent did, then improve it</h3>+
<p>Even with scoping, identity, and boundaries in place, an agent can respond in milliseconds, with no errors and every dashboard green, and still do the wrong thing. It may have retrieved the wrong information, chosen the wrong tool, or followed a path no one expected. A healthy system is not…+
<p><a href="https://aws.amazon.com/cloudwatch/omni/" target="_blank" rel="noopener">Amazon CloudWatch Omni</a> is built for continual observation. It captures every trace and lays out each step an agent took, from model calls to tool invocations, on a timeline a team can investigat…+
<h2 id="what-trust-in-depth-looks-like-in-practice"><strong>What Trust in Depth looks like in practice</strong></h2>+
<p>A manager asks a help desk agent to offboard Sam, an account manager who is leaving the company. The agent needs to remove Sam’s access, transfer his mailbox to his successor, and free up the license the company no longer needs. But Sam’s emails may also contain evidence needed for a client…+
<p>Two rules govern the sequence. Sam’s sign-in must be blocked before his mailbox can be converted and shared with his successor. Before changing his email license, the agent must check for a legal hold, a requirement to preserve records for a dispute, and confirm the change keeps those recor…+
<p><strong>Scope</strong>. Before the agent runs, the team writes these rules as policies in AgentCore. Each policy requires that the prerequisite step already succeeded for Sam’s account before the next step can run. The agent gets exactly six tools: block sign-in, check for legal…+
<p><strong>Know</strong>. The agent carries its own identity, so every change to Sam’s account traces back to it.</p>+
<p><strong>Bound</strong>. The agent tries to convert Sam’s mailbox to a shared one for Priya, the successor. Policy in AgentCore, enforced at the AgentCore Gateway, blocks the call: Sam can still sign in, and converting the mailbox wouldn’t change his password, so his old credenti…+
<p><strong>Observe</strong>. Every step is recorded: what the agent tried to do, what the rules decided, the hold, and the outcome.</p>+
<p>The agent completed the task without a person approving every step. Its permissions limited what it could change, its identity made those changes traceable, and policies stopped it from acting before the required safeguards were in place. The recorded steps let the team verify that Sam’s ac…+
<h2 id="what-becomes-possible-when-agents-are-trusted"><strong>What becomes possible when agents are trusted</strong></h2>+
<p>AWS brings these layers together so you can scope an agent to its job, know which agents are acting for you, bound what they do and say, see what they did and improve over time, and prove your rules hold wherever they can be stated precisely. No one layer earns that trust on its own. Togeth…+
<p>You draw the lines: what an agent may do on its own, what needs a person’s approval, and when it must stop. That frees your people for what only people can do: weigh the hard calls, resolve the gray areas, and own the decisions you choose to keep in human hands. As evidence builds that an a…+
<p>At AWS, we believe every organization deploying agents, on any platform, should be able to answer these four questions with evidence, and with proof wherever it applies. That is the bar AWS is building for our customers, and the one we invite the industry to hold itself to. Defense in Depth…+
<p>Start with one agent, one workflow, and one boundary you can prove. Then expand as the trust is earned.</p>+
<p>To begin implementing enforceable rules, explore <a href="https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/policy.html" target="_blank" rel="noopener">Policy in Amazon Bedrock AgentCore</a> and <a href="https://docs.aws.amazon.com/bedrock/latest/userguide/guardra…+
<p style="clear: both"></p>+
<hr style="width: 100%">+
<h2>About the author</h2>+
<footer>+
<div class="blog-author-box" style="padding-top: 2.0em">+
<div class="blog-author-image" style="margin-right: 1.0em">+
<img loading="lazy" class="alignnone size-full" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/14/ML-22119-2-1.jpg" alt="Shawn Bice" width="100" height="133">+
</div>+
<h3 class="lb-h4">Shawn Bice</h3>+
<p style="overflow: hidden">Shawn is Vice President of AI Services at Amazon Web Services (AWS), where he leads the Automated Reasoning Group. The team’s core mission is advancing AI agent safety. Shawn drives science and product teams to bring neurosymbolic AI, which combines large language…+
</div>+
</footer></content:encoded>+
+
+
+
</item>+
<item> <title>Build self-improving document processing using Amazon Bedrock AgentCore</title> <link>https://aws.amazon.com/blogs/machine-learning/build-self-improving-document-processing-using-amazon-bedrock-agentcore/</link> @
@@ -43,7 +332,7 @@<p>In this solution, eight specialized agents collaborate across two graph-based workflows running on Amazon Bedrock AgentCore.</p><p>The following diagram shows the multi-agent workflow graph built using <a href="https://strandsagents.com" target="_blank" rel="noopener">Strands Agents</a>, an open source SDK for building AI agents.</p><div id="attachment_141321" style="width: 1005px" class="wp-caption alignnone">-
<img aria-describedby="caption-attachment-141321" class="size-full wp-image-141321" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/14/Architecture-2.png" alt="" width="995" height="563">+
<img aria-describedby="caption-attachment-141321" loading="lazy" class="size-full wp-image-141321" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/14/Architecture-2.png" alt="" width="995" height="563"> <p id="caption-attachment-141321" class="wp-caption-text">Figure 1: Multi-agent orchestration workflow graph</p></div><ul>@
@@ -3985,202 +4274,6 @@ print(response)</code></pre> -
</item>-
<item>-
<title>Rethinking access control for RAG with Amazon Quick and Amazon Bedrock</title>-
<link>https://aws.amazon.com/blogs/machine-learning/rethinking-access-control-for-rag-with-amazon-quick-and-amazon-bedrock/</link>-
-
<dc:creator><![CDATA[Amit Choudhary]]></dc:creator>-
<pubDate>Wed, 07 Oct 2026 18:34:44 +0000</pubDate>-
<category><![CDATA[Advanced (300)]]></category>-
<category><![CDATA[Amazon Bedrock]]></category>-
<category><![CDATA[Amazon Quick Suite]]></category>-
<category><![CDATA[Announcements]]></category>-
<guid isPermaLink="false">39fc0b914168afce8435f252ea53c5d476f40b5d</guid>-
-
<description>Enterprise RAG unlocks insights from knowledge sources like SharePoint, Google Drive, and Confluence, but those sources carry complex permissions. Learn how Amazon Quick and Amazon Bedrock Knowledge Bases enforce document-level access controls in real time, verifying permissions di…-
<content:encoded><p>Enterprise organizations are adopting Retrieval Augmented Generation (RAG) to unlock insights from company knowledge sources like Microsoft SharePoint, Google Drive, and Atlassian Confluence. However, these knowledge sources contain sensitive information governed …-
<p>In this post, we explore how <a href="https://aws.amazon.com/quick/" target="_blank" rel="noopener">Amazon Quick</a> and <a href="https://aws.amazon.com/bedrock/knowledge-bases/" target="_blank" rel="noopener">Amazon Bedrock Knowledge Bases</a> solve this challenge t…-
<h2 id="the-business-problem">The business problem</h2>-
<p>Consider this scenario: A SharePoint site owner creates a knowledge base for their organization. Team members across multiple departments use an AI assistant to get answers from this knowledge base. The critical requirement is that each team member must only receive AI-generated insights fr…-
<p>This is a universal enterprise challenge. Organizations want to democratize access to AI-powered insights without compromising their existing security posture. A single unauthorized document surfaced in an AI response could expose confidential strategy documents, unreleased financial data, …-
<h2 id="why-existing-approaches-fall-short">Why existing approaches fall short</h2>-
<p>A common approach to RAG access control uses a replicate-and-filter approach to enforce document-level permissions. Here’s how it typically works:</p>-
<ol type="1">-
<li>A data source connector (for example, SharePoint connector) pulls ACLs as part of a periodic sync job.</li>-
<li>The ACLs are replicated from the data source and stored as attributes in an index.</li>-
<li>At query time, the AI system maps the logged-in user to the stored ACL attributes and filters results accordingly.</li>-
</ol>-
<p>While this approach seems reasonable on the surface, it has three fundamental weaknesses.</p>-
<h2 id="problem-1-the-ai-system-isnt-the-source-of-truth">Problem 1: The AI system isn’t the source of truth</h2>-
<p>In this model, the AI system takes sole responsibility for enforcement without being the authoritative source of permissions. This requires data connectors to accurately replicate complex, source-specific ACL logic across various data sources. Each data source has its own unique permission …-
<h2 id="problem-2-stale-permissions-create-security-gaps">Problem 2: Stale permissions create security gaps</h2>-
<p>In general, data connectors support pull-based syncs that run on demand or on a customer-defined schedule. The ACLs in these AI solutions are a snapshot in time from when the last sync ran. Some solutions use event-based updates, but this doesn’t work universally. For example, a data source…-
<h2 id="problem-3-evolving-data-source-capabilities">Problem 3: Evolving data source capabilities</h2>-
<p>Data sources regularly change or introduce new mechanisms to control access to content. A new permission feature in SharePoint or a change to the Google Drive sharing model could create gaps in the ACL mapping logic. This can expose content until the connector is updated.</p>-
<h2 id="how-aws-solves-this-real-time-acl-enforcement">How AWS solves this: Real-time ACL enforcement</h2>-
<p>To address these challenges, we implemented real-time ACL checks as an additional layer of security on top of existing pre-retrieval ACL filtering for Amazon Quick and Amazon Bedrock Knowledge Bases. This makes sure the system enforces the most current access controls by checking permission…-
<h2 id="architecture-overview">Architecture overview</h2>-
<p>The following diagram illustrates our hybrid approach that delivers both semantic search performance and real-time security capabilities.</p>-
<div style="width: 810px" class="wp-caption alignnone">-
<a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/07/ML-22080-1.png" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/07/ML-22080-1.png" alt="Two-stage ACL enforceme…-
<p class="wp-caption-text">Figure 1: Real-time ACL enforcement architecture for Amazon Quick and Amazon Bedrock Knowledge Bases, combining pre-retrieval filtering (Stage 1) with real-time verification against authoritative sources (Stage 2)</p>-
</div>-
<h2 id="how-it-works-a-google-drive-example">How it works: A Google Drive example</h2>-
<p>When a user submits a query to an Amazon Quick agent that uses a Google Drive knowledge base, the system enforces access controls in two stages:</p>-
<h3 id="stage-1-pre-retrieval-filtering">Stage 1: Pre-retrieval filtering</h3>-
<p>Amazon Quick performs a semantic search against the vector index to find the most relevant document passages. The system applies access control lists that are already stored in the index. This produces a preliminary set of candidate documents. This stage is necessary because real-time API c…-
<h3 id="stage-2-real-time-verification">Stage 2: Real-time verification</h3>-
<p>Amazon Quick verifies the candidate documents in real time by calling the Google Drive APIs. It uses the service account credential that the administrator provided to generate user-specific access tokens through impersonation. Google Drive maintains the source of truth for access control li…-
<p>This two-stage approach balances performance with security. It uses cached ACLs for efficiency while facilitating correctness through real-time checks. In addition to ACL enforcement, Amazon Bedrock provides responsible AI controls. These include Amazon Bedrock Guardrails for content filter…-
<h2 id="why-this-matters-for-your-organization">Why this matters for your organization</h2>-
<p>This approach delivers three key benefits:</p>-
<ul>-
<li><strong>Always-current permissions</strong> – No more security gaps between sync cycles when you are using a RAG product. If an employee’s access is revoked, the change is reflected in AI responses within moments, not hours or days.</li>-
<li><strong>Confidence to scale</strong> – Organizations can expand their knowledge base coverage knowing that real-time ACL checks verify permissions with the authoritative source for every query, regardless of data source.</li>-
<li><strong>Reduced operational burden</strong> – You don’t need to worry about sync frequency.</li>-
</ul>-
<h2 id="what-customers-are-saying-about-this">What customers are saying about this</h2>-
<blockquote>-
<p>“When we set out to evaluate AI solutions for our organization, our security and compliance teams were clear about their top priority: ensuring that colleagues would only ever see information they’re authorized to access. It’s a fundamental requirement, but one that many platforms struggle…-
<p>— Jamahl Wiggins, Sr.&nbsp;Specialist – M365 Innovation, Mondelēz International</p>-
</blockquote>-
<p>Mondelēz International has deployed Amazon Quick for their over 35,000 employees across four regions.</p>-
<h2 id="conclusion">Conclusion</h2>-
<p>In this post, we talked about how Amazon Quick and Amazon Bedrock Knowledge Bases implement real-time ACL enforcement to solve a critical security challenge for enterprises. The dual-layer ACL architecture verifies permissions directly with authoritative sources at query time. This makes su…-
<p>To get started, visit <a href="https://aws.amazon.com/quick/" target="_blank" rel="noopener">Amazon Quick</a> and <a href="https://aws.amazon.com/bedrock/knowledge-bases/" target="_blank" rel="noopener">Amazon Bedrock Knowledge Bases</a>.</p>-
<p style="clear: both"></p>-
<hr style="width: 100%">-
<h2>About the authors</h2>-
<footer>-
<div class="blog-author-box" style="padding-top: 2.0em">-
<div class="blog-author-image" style="margin-right: 1.0em">-
<img loading="lazy" class="alignnone size-full" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/07/ML-22080-2.jpg" alt="Amit Choudhary" width="100" height="133">-
</div>-
<h3 class="lb-h4">Amit Choudhary</h3>-
<p style="overflow: hidden">Amit is the Principal Product Manager for knowledge bases for Amazon Bedrock and Amazon Quick. His work enables secure AI interactions grounded in enterprise knowledge, transforming how organizations use their data for AI-powered insights and decision-making. He w…-
</div>-
<div class="blog-author-box" style="padding-top: 2.0em">-
<div class="blog-author-image" style="margin-right: 1.0em">-
<img loading="lazy" class="alignnone size-full" src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/07/ML-22080-3.jpg" alt="Suren Raju" width="100" height="133">-
</div>-
<h3 class="lb-h4">Suren Raju</h3>-
<p style="overflow: hidden">Suren is a Generative AI Specialist Solutions Architect at AWS on the Amazon Quick service team, where he helps enterprises build secure, agentic AI grounded in their own knowledge. He works at the frontier of agent-to-agent (A2A) orchestration and Model Context P…-
</div>Diff display stops at 400 lines. The line counts above are from the whole diff. 68 lines shown here cut at 300 characters. The raw artifact at this commit is linked above.