Change
5d56567
5d565670d76826a82f17bf3e445e457b0a3a1cfc · commit on GitHub
gcloud-blog-feed: changed (507760 bytes, HTTP 200)
raw/gcloud-blog-feed/response.xml modified
- Source
- gcloud-blog-feed
- Lines added
- +221
- Lines removed
- -324
- Stored bytes at this commit
- 507,760
- Timestamp
- observed
- Raw artifact at this commit
- raw/gcloud-blog-feed/response.xml
Recorded headers
| observed_at | 2026-10-21T04:47:22.360Z |
|---|---|
| origin_date | null |
| status | 200 |
| final URL | https://cloudblog.withgoogle.com/products/ai-machine-learning/rss/ |
| etag | null |
| last-modified | null |
| date | Wed, 21 Oct 2026 04:47:22 GMT |
| age | null |
| cache-control | public, max-age=600, s-maxage=600 |
| cf-cache-status | null |
| content-encoding | gzip |
| content-length | 133701 |
@
@@ -1,5 +1,58 @@<?xml version="1.0" encoding="utf-8"?>-
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:media="http://search.yahoo.com/mrss/"><channel><title>AI & Machine Learning</title><link>https://cloud.google.com/blog/products/a…+
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:media="http://search.yahoo.com/mrss/"><channel><title>AI & Machine Learning</title><link>https://cloud.google.com/blog/products/a…+
<div class="block-code"><dl>+
<dt>code_block</dt>+
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &#x27;---\r\nname: test-triager\r\ndescription: Runs unit tests, classifies failures, and returns a concise summary of failing assertions.\r\nmodel: flash\r\ntools:\r\n - read\r\n - edit\r\n---\r\n\r\nRun the requested t…+
</dl></div>+
<div class="block-paragraph"><p data-block-key="5p921">And a cross-model code reviewer subagent in <code>.agents/agents/gemini-judge.md</code>:</p></div>+
<div class="block-code"><dl>+
<dt>code_block</dt>+
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &quot;---\r\nname:cross-model-judge\r\ndescription: Independent LLM-as-a-judge verifier that audits diffs and edge cases using Gemini Pro.\r\nmodel: pro\r\ntools:\r\n - read\r\n---\r\n\r\nReview the current workspace diff…+
</dl></div>+
<div class="block-paragraph"><p data-block-key="5p921">With this setup, you can delegate rapid test-and-classify loops to <code>test-triager</code> and trigger <code>cross-model-judge</code> to validate the final diff, combining the strengths of each model in a si…+
<div class="block-video">+
+
+
+
<div class="article-module article-video ">+
<figure>+
<a class="h-c-video h-c-video--marquee"+
href="https://youtube.com/watch?v=ZapIlMXcBLo"+
data-glue-modal-trigger="uni-modal-ZapIlMXcBLo-"+
data-glue-modal-disabled-on-mobile="true">+
+
+
+
+
<div class="article-video__aspect-image"+
style="background-image: url(https://storage.googleapis.com/gweb-cloudblog-publish/images/63_iUmWoyf.max-1000x1000.png);">+
<span class="h-u-visually-hidden">Bringing Gemini to government services</span>+
</div>+
+
<svg role="img" class="h-c-video__play h-c-icon h-c-icon--color-white">+
<use xlink:href="#mi-youtube-icon"></use>+
</svg>+
</a>+
+
+
</figure>+
</div>+
+
<div class="h-c-modal--video"+
data-glue-modal="uni-modal-ZapIlMXcBLo-"+
data-glue-modal-close-label="Close Dialog">+
<a class="glue-yt-video"+
data-glue-yt-video-autoplay="true"+
data-glue-yt-video-height="99%"+
data-glue-yt-video-vid="ZapIlMXcBLo"+
data-glue-yt-video-width="100%"+
href="https://youtube.com/watch?v=ZapIlMXcBLo"+
ng-cloak>+
</a>+
</div>+
+
</div>+
<div class="block-paragraph"><p data-block-key="ecu27"><b>Five components of Gemini for government services</b></p><p data-block-key="cpbqg">Developed alongside agency leaders, Gemini for government services provides an integrated, fully governed environment confi…<p><span style="vertical-align: baseline;">But there’s still a big gap to be found in many automakers’ and Tier 1 suppliers’ cloud-based development processes. On the one hand, these companies are commonly using software to develop vehicles’ digital entertainment and information screens,…<p><span style="vertical-align: baseline;">On the other, the dozens of microchips and control modules that make a car a car these days — such as gateways, body controllers, zone control units, sensor and actuator nodes — are still frequently hardwired to physical chips when they’re const…<p><span style="vertical-align: baseline;">This inherent inefficiency has become a leading bottleneck. After all, manufacturers can’t really claim vehicles are software-defined if most of their electronics only function when connected to physical hardware and if developers are leaning he…@
@@ -89,7 +142,162 @@</div>-
<div class="block-paragraph"><h3 data-block-key="3itrx">Gemini meets work where it already happens</h3><p data-block-key="7fq23">A viable agentic platform has to meet work where it already happens – and run under production-grade controls. Gemini is engineered to deliver deve…+
<div class="block-paragraph"><h3 data-block-key="3itrx">Gemini meets work where it already happens</h3><p data-block-key="7fq23">A viable agentic platform has to meet work where it already happens – and run under production-grade controls. Gemini is engineered to deliver deve…+
<p><span style="vertical-align: baseline;">Evaluation engineering is a critical foundation for building reliable AI systems. Setting up systematic, repeatable tests to measure your agent’s true behavior can be the difference between shipping a tool you can trust and one that silently bre…+
<p><span style="vertical-align: baseline;">I’ve run evaluation engineering initiatives for years at Google and d</span><span style="vertical-align: baseline;">eveloping the newly released</span><a href="https://cloud.google.com/blog/topics/developers-practitioners/in…+
<p><span style="vertical-align: baseline;">Here are four eval lessons we learned from building the new plugin that you can apply to your own builds.</span></p>+
<h3><span style="vertical-align: baseline;">1. Grade the execution steps instead of the summary</span></h3>+
<p><span style="vertical-align: baseline;">When we first evaluated agents, we set up an LLM-as-judge to inspect the agent’s final text reply and scored it against a checklist. However, we quickly realized it’s fundamentally flawed because the grader only sees what the agent wrote, not wh…+
<div class="block-image_full_width">+
+
+
+
+
+
+
+
<div class="article-module h-c-page">+
<div class="h-c-grid">+
+
+
<figure class="article-image--large+
+
+
h-c-grid__col+
h-c-grid__col--6 h-c-grid__col--offset-3+
+
+
"+
>+
+
+
+
+
<img+
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/Image_1_XmJ1pt2.max-1000x1000.png"+
+
alt="Image 1">+
+
</a>+
+
</figure>+
+
+
</div>+
</div>+
+
+
+
+
+
</div>+
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The grader penalized the agent for doing the right thing simply because it didn’t brag about it in the summary. If your rubric line says "did it run X?", a LLM judge is guessing unless you evaluate the…+
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">:</strong><span style="vertical-align: baseline;"> Don't grade an agent on its closing pitch. Score the raw execution trajectory to verify what commands were ac…+
<h3><span style="vertical-align: baseline;">2. Design evals for asynchronous operations</span></h3>+
<p><span style="vertical-align: baseline;">When you start reading tool-call trajectories, how you extract that telemetry matters just as much as what you're reading. Extracting agent telemetry into a flattened, sequential list of events obscures concurrent tool dispatch, which is one of …+
<p><span style="vertical-align: baseline;">In our evaluations, when an agent was asked to inspect an environment and look up documentation, the harness dispatched the skill-loading call and two documentation searches simultaneously within a single tool_calls payload. Because activating a…+
<div class="block-image_full_width">+
+
+
+
+
+
+
+
<div class="article-module h-c-page">+
<div class="h-c-grid">+
+
+
<figure class="article-image--large+
+
+
h-c-grid__col+
h-c-grid__col--6 h-c-grid__col--offset-3+
+
+
"+
>+
+
+
+
+
<img+
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/image_2_CHyk3qC.max-1000x1000.png"+
+
alt="image 2">+
+
</a>+
+
</figure>+
+
+
</div>+
</div>+
+
+
+
+
+
</div>+
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The system never gave the skill a dedicated turn to let its guidance influence the agent's actions. If telemetry pipelines strip out turn structures and merge calls into a flat timeline, engineers will…+
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">: </strong><span style="vertical-align: baseline;">Preserve turn structure in telemetry extractors so you can identify when client-side parallel dispatch issues…+
<h3><span style="vertical-align: baseline;">3. Test live runs and "dry runs" side-by-side</span></h3>+
<p><span style="vertical-align: baseline;">You need to inspect real API calls to catch harness quirks, like parallel dispatch. But if you only test your agent in one environment condition, you're flying half-blind. From day one, you need to run two conditions side-by-side: a live run and…+
<p><span style="vertical-align: baseline;">Live runs frequently hit an unexpected permission error on step two and abort before you ever reach the behavior you’re trying to test. That's why you need the plan-only condition, where you strip credentials and tell the agent to explain what i…+
<div class="block-image_full_width">+
+
+
+
+
+
+
+
<div class="article-module h-c-page">+
<div class="h-c-grid">+
+
+
<figure class="article-image--large+
+
+
h-c-grid__col+
h-c-grid__col--6 h-c-grid__col--offset-3+
+
+
"+
>+
+
+
+
+
<img+
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/image_3_OixTmu5.max-1000x1000.png"+
+
alt="image 3">+
+
</a>+
+
</figure>+
+
+
</div>+
</div>+
+
+
+
+
+
</div>+
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">For example, on one of our onboarding prompts, a live run was blocked immediately on a project list call. The plan-only run sailed right through and produced an 8,000-character plan that completely ign…+
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">:</strong><span style="vertical-align: baseline;"> Run live and plan-only evaluations side by side. Live runs test actual execution and safety restraint; plan-o…+
<h3><span style="vertical-align: baseline;">4. Use “golden sets” as durable baselines</span></h3>+
<p><span style="vertical-align: baseline;">Frontier model architectures change every few months, agent harnesses improve, and plugin manifest standards will inevitably evolve. When building the Google Cloud Developer plugin, we treated evaluations as continuous integration testing for th…+
<p><span style="vertical-align: baseline;">Instead of guessing from the final text output, this approach captures the whole trajectory: what the agent thought, what commands it actually ran, and whether it solved the problem without breaking anything. Then you score that run against gold…+
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">: </strong><span style="vertical-align: baseline;">Build golden sets based on what the entire system should actually do in the wild, including your harness, mod…+
<p><strong style="vertical-align: baseline;">Reminder: Use evaluation suites over “one-shot” demos</strong></p>+
<p><span style="vertical-align: baseline;">A common mistake I see among engineers is what I like to call the one-shot demo. This is when someone feeds a prompt into a LLM, watches it build a complete 3D game or solves a Rubik’s cube in a single pass and calls it an eval. But in reality, …+
<p><span style="vertical-align: baseline;">To do quality work with AI, I truly believe that you need evaluation suites. Without evaluations, teams are like literally running into a dark cave blind. Engineering an enterprise-grade agent means testing the system's ability to handle multi-s…+
<h3><span style="vertical-align: baseline;">Where to go from here</span></h3>+
<p><span style="vertical-align: baseline;">Building an agent plugin or MCP server is the beginning. Shipping something you can actually trust with real cloud infrastructure requires significantly more work in evaluations to ensure it’s doing what you expect it to do. At the end of the da…<p style="text-align: justify;"><span style="vertical-align: baseline;">Folding a protein target, screening a 50,000 compound ligand library, or evaluating structure-activity relationships (SAR) are not quick API lookups. They are sustained, stateful, GPU-bound simulations that run for m…<p style="text-align: justify;"><span style="vertical-align: baseline;">Building enterprise grade research and development (R&amp;D) agents requires moving past the monolithic assistant model. In this blog post, we will discuss how to use a three-layer architecture: separating cognit…<div class="block-video">@
@@ -227,6 +435,9 @@<div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;"><div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;"><div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;">+
<div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;">+
<div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;">+
<div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;"><div style="color: #5f6368; overflow-x: auto; overflow-y: hidden; width: 100%;"><table><colgroup><col/><col/><col/></colgroup><tbody><tr>@
@@ -308,6 +519,9 @@</div></div></div>+
</div>+
</div>+
</div><h4><span style="vertical-align: baseline;">The enterprise governance and asymmetric autonomy</span></h4><p style="text-align: justify;"><span style="vertical-align: baseline;">Deploying high-compute agents in enterprise environments requires robust data sovereignty, audit trails, and process control. The </span><a href="https://cloud.google.com/blog/products/ai-machine-learning/in…<ul>@
@@ -342,162 +556,7 @@</li></ol><p style="text-align: justify;"><span style="vertical-align: baseline;">As generative and structural biology models expand toward whole-cell modeling and complex dynamic simulations, this decoupled architecture ensures research systems scale efficiently while keeping proprietary IP secur…-
<p style="text-align: justify;"><span style="vertical-align: baseline;">Reach out to your Google sales representative if you have any requirements to build a use case like this and we will be happy to talk and support you in your agent development journey on Google Cloud.</span><…-
<p><span style="vertical-align: baseline;">Evaluation engineering is a critical foundation for building reliable AI systems. Setting up systematic, repeatable tests to measure your agent’s true behavior can be the difference between shipping a tool you can trust and one that silently bre…-
<p><span style="vertical-align: baseline;">I’ve run evaluation engineering initiatives for years at Google and d</span><span style="vertical-align: baseline;">eveloping the newly released</span><a href="https://cloud.google.com/blog/topics/developers-practitioners/in…-
<p><span style="vertical-align: baseline;">Here are four eval lessons we learned from building the new plugin that you can apply to your own builds.</span></p>-
<h3><span style="vertical-align: baseline;">1. Grade the execution steps instead of the summary</span></h3>-
<p><span style="vertical-align: baseline;">When we first evaluated agents, we set up an LLM-as-judge to inspect the agent’s final text reply and scored it against a checklist. However, we quickly realized it’s fundamentally flawed because the grader only sees what the agent wrote, not wh…-
<div class="block-image_full_width">-
-
-
-
-
-
-
-
<div class="article-module h-c-page">-
<div class="h-c-grid">-
-
-
<figure class="article-image--large-
-
-
h-c-grid__col-
h-c-grid__col--6 h-c-grid__col--offset-3-
-
-
"-
>-
-
-
-
-
<img-
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/Image_1_XmJ1pt2.max-1000x1000.png"-
-
alt="Image 1">-
-
</a>-
-
</figure>-
-
-
</div>-
</div>-
-
-
-
-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The grader penalized the agent for doing the right thing simply because it didn’t brag about it in the summary. If your rubric line says "did it run X?", a LLM judge is guessing unless you evaluate the…-
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">:</strong><span style="vertical-align: baseline;"> Don't grade an agent on its closing pitch. Score the raw execution trajectory to verify what commands were ac…-
<h3><span style="vertical-align: baseline;">2. Design evals for asynchronous operations</span></h3>-
<p><span style="vertical-align: baseline;">When you start reading tool-call trajectories, how you extract that telemetry matters just as much as what you're reading. Extracting agent telemetry into a flattened, sequential list of events obscures concurrent tool dispatch, which is one of …-
<p><span style="vertical-align: baseline;">In our evaluations, when an agent was asked to inspect an environment and look up documentation, the harness dispatched the skill-loading call and two documentation searches simultaneously within a single tool_calls payload. Because activating a…-
<div class="block-image_full_width">-
-
-
-
-
-
-
-
<div class="article-module h-c-page">-
<div class="h-c-grid">-
-
-
<figure class="article-image--large-
-
-
h-c-grid__col-
h-c-grid__col--6 h-c-grid__col--offset-3-
-
-
"-
>-
-
-
-
-
<img-
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/image_2_CHyk3qC.max-1000x1000.png"-
-
alt="image 2">-
-
</a>-
-
</figure>-
-
-
</div>-
</div>-
-
-
-
-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The system never gave the skill a dedicated turn to let its guidance influence the agent's actions. If telemetry pipelines strip out turn structures and merge calls into a flat timeline, engineers will…-
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">: </strong><span style="vertical-align: baseline;">Preserve turn structure in telemetry extractors so you can identify when client-side parallel dispatch issues…-
<h3><span style="vertical-align: baseline;">3. Test live runs and "dry runs" side-by-side</span></h3>-
<p><span style="vertical-align: baseline;">You need to inspect real API calls to catch harness quirks, like parallel dispatch. But if you only test your agent in one environment condition, you're flying half-blind. From day one, you need to run two conditions side-by-side: a live run and…-
<p><span style="vertical-align: baseline;">Live runs frequently hit an unexpected permission error on step two and abort before you ever reach the behavior you’re trying to test. That's why you need the plan-only condition, where you strip credentials and tell the agent to explain what i…-
<div class="block-image_full_width">-
-
-
-
-
-
-
-
<div class="article-module h-c-page">-
<div class="h-c-grid">-
-
-
<figure class="article-image--large-
-
-
h-c-grid__col-
h-c-grid__col--6 h-c-grid__col--offset-3-
-
-
"-
>-
-
-
-
-
<img-
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/image_3_OixTmu5.max-1000x1000.png"-
-
alt="image 3">-
-
</a>-
-
</figure>-
-
-
</div>-
</div>-
-
-
-
-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">For example, on one of our onboarding prompts, a live run was blocked immediately on a project list call. The plan-only run sailed right through and produced an 8,000-character plan that completely ign…-
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">:</strong><span style="vertical-align: baseline;"> Run live and plan-only evaluations side by side. Live runs test actual execution and safety restraint; plan-o…-
<h3><span style="vertical-align: baseline;">4. Use “golden sets” as durable baselines</span></h3>-
<p><span style="vertical-align: baseline;">Frontier model architectures change every few months, agent harnesses improve, and plugin manifest standards will inevitably evolve. When building the Google Cloud Developer plugin, we treated evaluations as continuous integration testing for th…-
<p><span style="vertical-align: baseline;">Instead of guessing from the final text output, this approach captures the whole trajectory: what the agent thought, what commands it actually ran, and whether it solved the problem without breaking anything. Then you score that run against gold…-
<p><strong style="vertical-align: baseline;">Takeaway</strong><strong style="vertical-align: baseline;">: </strong><span style="vertical-align: baseline;">Build golden sets based on what the entire system should actually do in the wild, including your harness, mod…-
<p><strong style="vertical-align: baseline;">Reminder: Use evaluation suites over “one-shot” demos</strong></p>Diff display stops at 400 lines. The line counts above are from the whole diff. 51 lines shown here cut at 300 characters. The raw artifact at this commit is linked above.