Change
ebe5996
ebe599660f5cd125087176d3ca9845c3ac25a518 · commit on GitHub
gcloud-blog-feed: changed (505362 bytes, HTTP 200)
raw/gcloud-blog-feed/response.xml modified
- Source
- gcloud-blog-feed
- Lines added
- +162
- Lines removed
- -512
- Stored bytes at this commit
- 505,362
- Timestamp
- observed
- Raw artifact at this commit
- raw/gcloud-blog-feed/response.xml
Recorded headers
| observed_at | 2026-10-20T04:51:24.749Z |
|---|---|
| origin_date | null |
| status | 200 |
| final URL | https://cloudblog.withgoogle.com/products/ai-machine-learning/rss/ |
| etag | null |
| last-modified | null |
| date | Tue, 20 Oct 2026 04:51:24 GMT |
| age | null |
| cache-control | public, max-age=600, s-maxage=600 |
| cf-cache-status | null |
| content-encoding | gzip |
| content-length | 130670 |
@
@@ -1,5 +1,95 @@<?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…+
<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…+
<p><span style="vertical-align: baseline;">With Horizon, our agentic innovation platform for developing software-defined vehicles, our vision is to help manufacturers avoid these inefficiencies by working with partners like </span><a href="https://www.remotivelabs.com/" rel="noo…+
<h3><strong style="vertical-align: baseline;">The cockpit already runs in the cloud</strong></h3>+
<p><span style="vertical-align: baseline;">The Android team has invested a significant amount of time and effort over the last few years on enabling seamless development of the many digital systems in vehicles’ cockpits. </span><a href="https://source.android.com/docs/automotive…+
<p><span style="vertical-align: baseline;">And with the recent launch of </span><a href="https://source.android.com/docs/whatsnew/android-17-release#sdv" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Android 17 and AA…+
<h3><strong style="vertical-align: baseline;">The harness around the virtual vehicle</strong></h3>+
<p><span style="vertical-align: baseline;">That’s where </span><a href="https://www.remotivelabs.com/products/remotivetopology" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">RemotiveTopology</span></a><…+
<p><span style="vertical-align: baseline;">RemotiveTopology automatically utilizes manufacturers’ existing network design files to build a working, virtual version of their cars’ networks. This is vehicle-as-code: the E/E architecture expressed as an executable specification, versioned a…+
<h3><strong style="vertical-align: baseline;">Integrate across domains early, not late</strong></h3>+
<p><span style="vertical-align: baseline;">The payoff for using this approach is that all onboard systems can be developed and integrated with software before any of them are connected to a vehicle’s hardware. That’s beneficial because the most difficult (and latest) integration in a veh…+
<p><span style="vertical-align: baseline;">With the full topology live on Google Cloud, the infotainment system and ADAS can operate simultaneously within a virtual vehicle, exchanging real signals over the simulated network. This means it’s possible to exercise handoffs like an ADAS eve…+
<h3><strong style="vertical-align: baseline;">The full vehicle frontier</strong></h3>+
<p><span style="vertical-align: baseline;">At the heart of this work is a shared desire to bring scalable software development to even the smallest and most common nodes in a vehicle, its microchips and control modules. That’s because these are the parts that have traditionally been wire…+
<p><span style="vertical-align: baseline;">The ultimate goal is that the same software that runs in the cloud can run on vehicles’ actual control modules with no re-tooling or re-validation needed to move between them. That parity is natural with Arm-based Google Cloud infrastructure bec…+
<p><span style="vertical-align: baseline;">Taken together, this holistic system allows teams, for the first time, to develop and integration-test an entire vehicle’s systems in the cloud and then apply it to the actual vehicle without having to rewrite it.</span></p>+
<h3><strong style="vertical-align: baseline;">Why now, and where to start</strong></h3>+
<p><span style="vertical-align: baseline;">The payoff is not abstract. Treating the vehicle as code and moving integration into the cloud compresses the slowest, most expensive phase of a program. Early customer engagements with RemotiveLabs have measured more than 10x faster integration…+
<p><span style="vertical-align: baseline;">RemotiveTopology provides this architecture as vehicle-as-code and makes it accessible for developers, testers, and autonomous AI agents alike. </span><span style="vertical-align: baseline;">Best of all, there is no new third-party t…+
<p><span style="vertical-align: baseline;">For those building software-defined vehicles whose virtual development stops at the cockpit, it’s time to start with an open-source example, bring unique platform definitions, and run an entire vehicle in the cloud. </span><a href="http…+
<p style="text-align: justify;"><span style="vertical-align: baseline;">There’s a common misconception in tech that humans are subjective, so we should rely on AI for objective evaluation. But the reality is that LLMs are also subjective. Just like human reviewers, they can carry biases:…+
<ul>+
<li role="presentation"><strong style="vertical-align: baseline;">Verbosity bias</strong><span style="vertical-align: baseline;">: LLM judges usually favor </span><a href="https://arxiv.org/abs/2411.15594" rel="noopener" target="_blank"><span style="text-decora…+
<li role="presentation"><strong style="vertical-align: baseline;">Position bias</strong><span style="vertical-align: baseline;">: Swapping the order of two options can cause an LLM judge to flip its verdict </span><a href="https://arxiv.org/abs/2306.05685" rel="noope…+
<li role="presentation"><strong style="vertical-align: baseline;">The halo effect</strong><span style="vertical-align: baseline;">: Because LLMs generate text piece by piece, </span><a href="https://arxiv.org/abs/2411.16337" rel="noopener" target="_blank"><span…+
<li role="presentation"><strong style="vertical-align: baseline;">Agreeableness bias</strong><span style="vertical-align: baseline;">: LLM judges are great at recognizing valid work. But with detecting errors, they may fail to catch incorrect, buggy, or unsafe outputs </sp…+
<li role="presentation"><strong style="vertical-align: baseline;">Self-enhancement bias:</strong><span style="vertical-align: baseline;"> Also known as egocentric bias, AI evaluators systematically </span><a href="https://arxiv.org/abs/2309.17012" rel="noopener" targ…+
</ul>+
<p style="text-align: justify;"><span style="vertical-align: baseline;">Ultimately, LLM judges are not objective arbiters of truth. They are subjective models with their own distinct biases. Treating them as infallible tools that don't need their quality measured might leave your work co…+
<p style="text-align: justify;"><span style="vertical-align: baseline;">In this post, we’ll explore what’s holding enterprises back from strong evals, and share an easy three-pillar method you can use to improve your quality. </span></p>+
<h3 style="text-align: justify;"><span style="vertical-align: baseline;">Three helpful ways to audit your LLM judges </span></h3>+
<p><strong style="vertical-align: baseline;">Pillar 1: Measure internal stability</strong></p>+
<p style="text-align: justify;"><span style="vertical-align: baseline;">Before you can trust an AI judge’s score, you need to know that it isn't just guessing. Because LLMs are probabilistic, they can be inconsistent, giving different scores to the exact same text.</span></p>+
<p style="text-align: justify;"><span style="vertical-align: baseline;">You can test this internal stability by running multi-pass consistency checks on a single output and calculating the variance. If a judge's scores fluctuate wildly on identical inputs, its rubric is too ambiguous.<…+
<p><strong style="vertical-align: baseline;">Pillar 2: Validate human alignment</strong></p>+
<p style="text-align: justify;"><span style="vertical-align: baseline;">When your evaluation datasets are mostly valid responses, a lenient AI judge can look great on paper. It might align closely with human reviewers overall, but miss a majority of invalid or even unsafe outputs.</sp…+
<p style="text-align: justify;"><span style="vertical-align: baseline;">To prove your LLM judge is catching real errors, you need chance-corrected metrics. These mathematical models filter out agreements driven purely by leniency or random chance, showing how closely your evaluator actua…+
<p><strong style="vertical-align: baseline;">Pillar 3: Use multi-judge ensembles</strong></p>+
<p style="text-align: justify;"><span style="vertical-align: baseline;">To avoid the halo effect mentioned earlier, teams should use decomposed prompting. Instead of asking one LLM judge to score an entire response, assign a specialized judge to evaluate a single criteria, like factual a…+
<p style="text-align: justify;"><span style="vertical-align: baseline;">Second, to track performance in real time without human intervention, the established best practice is ensembling. This means running at least two different models under the hood to evaluate the exact same output sim…+
<p style="text-align: justify;"><span style="vertical-align: baseline;">By finding the statistical consensus between these models, you can flag if their evaluations begin to diverge and avoid any scoring quality degradation.</span></p>+
<h3 style="text-align: justify;"><span style="vertical-align: baseline;">Get started</span></h3>+
<p style="text-align: justify;"><a href="https://cloud.google.com/consulting/innovation-and-transformation?e=48754805"><span style="text-decoration: underline; vertical-align: baseline;">Partner with Delta</span></a><span style="vertical-align: baseline;"> to desi…+
<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_5_i5rXZEH.max-1000x1000.png"+
+
alt="IDC image v2">+
+
</a>+
+
<figcaption class="article-image__caption "><p data-block-key="huudm">IDC MarketScape vendor analysis model is designed to provide an overview of the competitive fitness of technology and suppliers in a given market. The research methodology utilizes a rigorous scoring methodolog…+
+
</figure>+
+
+
</div>+
</div>+
+
+
+
+
+
</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…<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">@
@@ -407,72 +497,7 @@<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><span style="vertical-align: baseline;">Last week at Gemini at Work 2026, </span><a href="https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026"><span style="text-decoration: underline; vertical-align: baseline;">we introduced the …-
<p><span style="vertical-align: baseline;">Here is a look at how the Gemini agent works, five practical takeaways for founders, and how startups are already using it to punch above their weight.</span></p>-
<h3><strong style="vertical-align: baseline;">What is the Gemini agent?</strong></h3>-
<p><span style="vertical-align: baseline;">Instead of prompting separate chatbots for writing, coding, and data analysis, the Gemini agent is a single agent that takes high-level objectives and returns finished work. Three principles define it:</span></p>-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Unified:</strong><span style="vertical-align: baseline;"> A single agent can now handle knowledge work, data analysis, creative asset generation, and writing and running code. For complex jobs, it dynamically sp…-
</li>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Omnipresent access:</strong><span style="vertical-align: baseline;"> It works across the surfaces you use every day — on the web, command line, desktop, and mobile, as well as natively inside Gmail, Docs, Sheets…-
</li>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Persistent:</strong><span style="vertical-align: baseline;"> It runs in the cloud with continuous memory across four dimensions (session, semantic, procedural, and episodic). It onboards itself like a new hire —…-
</li>-
</ul>-
<p><span style="vertical-align: baseline;">For founders wearing five hats at once, Gemini works like a digital chief of staff that's already briefed on your calendar, projects and documents. Ask it to research market trends, build a financial model in Sheets and turn both into a pitch de…-
<h3><strong style="vertical-align: baseline;">Five takeaways for startups</strong></h3>-
<p><strong style="vertical-align: baseline;">1. Coworker agents and skills help you scale your output</strong></p>-
<p><span style="vertical-align: baseline;">Beyond personal assistance, you can create persistent </span><strong style="vertical-align: baseline;">coworker agents</strong><span style="vertical-align: baseline;"> that work with your entire team. Each agent comes wit…-
<p><span style="vertical-align: baseline;">You can also equip agents with reusable </span><strong style="vertical-align: baseline;">skills</strong><span style="vertical-align: baseline;"> — or let Gemini write skills for itself as it learns your processes. For exa…-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> Coworker agents can immediately increase the capacity of your teams, and skills can standardize repeatable processes like investor and…-
</li>-
</ul>-
<p><strong style="vertical-align: baseline;">2. Open and third-party model choice</strong></p>-
<p><span style="vertical-align: baseline;">Gemini is the agent, and the model underneath is your choice. As startup architectures mature, sending every task to a single frontier model quickly eats into gross margins. With the Gemini agent, you can dynamically route workloads across Gemin…-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> You get flexibility without lock-in. You can optimize your cost per task — using compact models for high-volume, deterministic jobs an…-
</li>-
</ul>-
<p><strong style="vertical-align: baseline;">3. Connected to your services and tools </strong></p>-
<p><span style="vertical-align: baseline;">An agent is only as useful as the systems it can reach. In addition to Google Workspace, the Gemini agent connects directly to your engineering, data, and business stack, including partner solutions:</span></p>-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Development and collaboration:</strong><span style="vertical-align: baseline;"> Git, Slack, Jira and more </span></p>-
</li>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Data and local files:</strong><span style="vertical-align: baseline;"> BigQuery, Postgres, Snowflake, Databricks, and more alongside files on your desktop</span></p>-
</li>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Custom tools via MCP:</strong><span style="vertical-align: baseline;"> Any Model Context Protocol (MCP) server inside or outside your network so your engineers can build and share internal tools.</span><…-
</li>-
</ul>-
<p><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> Instead of spending engineering cycles maintaining custom connectors, your team can plug the agent straight into your existing codebase, databases, and MC…-
<p><strong style="vertical-align: baseline;">4. Built-in security and governance</strong></p>-
<p><span style="vertical-align: baseline;">For B2B startups, the enterprise security audit is often where deals stall. Under the hood, every agent is assigned a cryptographic </span><strong style="vertical-align: baseline;">Agent Identity</strong><span style="vertica…-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> When an enterprise buyer’s CISO asks how your AI workflows handle data privacy, prompt injection, and access control, you have an ente…-
</li>-
</ul>-
<p><strong style="vertical-align: baseline;">5. Protect your runway with real-time cost controls</strong></p>-
<p><span style="vertical-align: baseline;">Autonomous agents that run sub-agents for hours or days can quickly run up a bill if a loop goes unchecked. You can now set hard, real-time spend caps directly in the Cloud Billing Console; if a cap is reached, the project's agent automatically …-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> You and your developers can experiment with long-running, parallel agentic loops with a built-in financial safety net that protects yo…-
</li>-
</ul>-
<h3><strong style="vertical-align: baseline;">Founders already building</strong></h3>-
<p><span style="vertical-align: baseline;">Startups and AI-native innovators are already putting Gemini to work. Immersive storytelling platform</span><a href="https://cloud.google.com/transform/whenere-dreamlands-gemini-jane-austen-fortnite-future-of-fan-fiction"><span st…-
<h3><strong style="vertical-align: baseline;">Build with Google for Startups</strong></h3>-
<p><span style="vertical-align: baseline;">The agentic era gives lean teams the leverage to punch far above their weight — whether you’re deploying coworker agents to cover internal operations or building agentic workflows directly into your product.</span></p>-
<p><span style="vertical-align: baseline;">While these new Gemini agent capabilities are not publicly available yet, startups can</span><a href="https://cloud.google.com/contact"><span style="vertical-align: baseline;"> </span><span style="text-decoration: unde…+
<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…<hr/><p><span style="vertical-align: baseline;">What makes a live video world model so challenging to serve in production? The bottleneck isn't average throughput — it's beating a relentless real-time clock. Unlike offline video generators where a user submits a prompt and waits in a queue, a…<p><span style="vertical-align: baseline;">Today, we're excited to share how </span><a href="https://visko.ai" rel="noopener" target="_blank"><strong style="text-decoration: underline; vertical-align: baseline;">Visko</strong></a><span style="vertical-ali…@
@@ -684,7 +709,72 @@</li></ul><hr/>-
<p><sup><span style="font-style: italic; vertical-align: baseline;">Special thanks to the Visko.ai team for the close technical partnership, and to the Google Cloud TPU engineers who lead the project including: Yixin Shi, James Huang, and Elisa Tsai.</span></sup></p&…+
<p><sup><span style="font-style: italic; vertical-align: baseline;">Special thanks to the Visko.ai team for the close technical partnership, and to the Google Cloud TPU engineers who lead the project including: Yixin Shi, James Huang, and Elisa Tsai.</span></sup></p&…+
<p><span style="vertical-align: baseline;">Last week at Gemini at Work 2026, </span><a href="https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026"><span style="text-decoration: underline; vertical-align: baseline;">we introduced the …+
<p><span style="vertical-align: baseline;">Here is a look at how the Gemini agent works, five practical takeaways for founders, and how startups are already using it to punch above their weight.</span></p>+
<h3><strong style="vertical-align: baseline;">What is the Gemini agent?</strong></h3>+
<p><span style="vertical-align: baseline;">Instead of prompting separate chatbots for writing, coding, and data analysis, the Gemini agent is a single agent that takes high-level objectives and returns finished work. Three principles define it:</span></p>+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Unified:</strong><span style="vertical-align: baseline;"> A single agent can now handle knowledge work, data analysis, creative asset generation, and writing and running code. For complex jobs, it dynamically sp…+
</li>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Omnipresent access:</strong><span style="vertical-align: baseline;"> It works across the surfaces you use every day — on the web, command line, desktop, and mobile, as well as natively inside Gmail, Docs, Sheets…+
</li>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Persistent:</strong><span style="vertical-align: baseline;"> It runs in the cloud with continuous memory across four dimensions (session, semantic, procedural, and episodic). It onboards itself like a new hire —…+
</li>+
</ul>+
<p><span style="vertical-align: baseline;">For founders wearing five hats at once, Gemini works like a digital chief of staff that's already briefed on your calendar, projects and documents. Ask it to research market trends, build a financial model in Sheets and turn both into a pitch de…+
<h3><strong style="vertical-align: baseline;">Five takeaways for startups</strong></h3>+
<p><strong style="vertical-align: baseline;">1. Coworker agents and skills help you scale your output</strong></p>+
<p><span style="vertical-align: baseline;">Beyond personal assistance, you can create persistent </span><strong style="vertical-align: baseline;">coworker agents</strong><span style="vertical-align: baseline;"> that work with your entire team. Each agent comes wit…+
<p><span style="vertical-align: baseline;">You can also equip agents with reusable </span><strong style="vertical-align: baseline;">skills</strong><span style="vertical-align: baseline;"> — or let Gemini write skills for itself as it learns your processes. For exa…+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> Coworker agents can immediately increase the capacity of your teams, and skills can standardize repeatable processes like investor and…+
</li>+
</ul>+
<p><strong style="vertical-align: baseline;">2. Open and third-party model choice</strong></p>+
<p><span style="vertical-align: baseline;">Gemini is the agent, and the model underneath is your choice. As startup architectures mature, sending every task to a single frontier model quickly eats into gross margins. With the Gemini agent, you can dynamically route workloads across Gemin…+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> You get flexibility without lock-in. You can optimize your cost per task — using compact models for high-volume, deterministic jobs an…+
</li>+
</ul>+
<p><strong style="vertical-align: baseline;">3. Connected to your services and tools </strong></p>+
<p><span style="vertical-align: baseline;">An agent is only as useful as the systems it can reach. In addition to Google Workspace, the Gemini agent connects directly to your engineering, data, and business stack, including partner solutions:</span></p>+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Development and collaboration:</strong><span style="vertical-align: baseline;"> Git, Slack, Jira and more </span></p>+
</li>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Data and local files:</strong><span style="vertical-align: baseline;"> BigQuery, Postgres, Snowflake, Databricks, and more alongside files on your desktop</span></p>+
</li>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">Custom tools via MCP:</strong><span style="vertical-align: baseline;"> Any Model Context Protocol (MCP) server inside or outside your network so your engineers can build and share internal tools.</span><…+
</li>+
</ul>+
<p><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> Instead of spending engineering cycles maintaining custom connectors, your team can plug the agent straight into your existing codebase, databases, and MC…+
<p><strong style="vertical-align: baseline;">4. Built-in security and governance</strong></p>+
<p><span style="vertical-align: baseline;">For B2B startups, the enterprise security audit is often where deals stall. Under the hood, every agent is assigned a cryptographic </span><strong style="vertical-align: baseline;">Agent Identity</strong><span style="vertica…+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> When an enterprise buyer’s CISO asks how your AI workflows handle data privacy, prompt injection, and access control, you have an ente…+
</li>+
</ul>+
<p><strong style="vertical-align: baseline;">5. Protect your runway with real-time cost controls</strong></p>+
<p><span style="vertical-align: baseline;">Autonomous agents that run sub-agents for hours or days can quickly run up a bill if a loop goes unchecked. You can now set hard, real-time spend caps directly in the Cloud Billing Console; if a cap is reached, the project's agent automatically …+
<ul>+
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">+
<p role="presentation"><strong style="vertical-align: baseline;">What it means for startups:</strong><span style="vertical-align: baseline;"> You and your developers can experiment with long-running, parallel agentic loops with a built-in financial safety net that protects yo…+
</li>+
</ul>+
<h3><strong style="vertical-align: baseline;">Founders already building</strong></h3>+
<p><span style="vertical-align: baseline;">Startups and AI-native innovators are already putting Gemini to work. Immersive storytelling platform</span><a href="https://cloud.google.com/transform/whenere-dreamlands-gemini-jane-austen-fortnite-future-of-fan-fiction"><span st…+
<h3><strong style="vertical-align: baseline;">Build with Google for Startups</strong></h3>+
<p><span style="vertical-align: baseline;">The agentic era gives lean teams the leverage to punch far above their weight — whether you’re deploying coworker agents to cover internal operations or building agentic workflows directly into your product.</span></p>+
<p><span style="vertical-align: baseline;">While these new Gemini agent capabilities are not publicly available yet, startups can</span><a href="https://cloud.google.com/contact"><span style="vertical-align: baseline;"> </span><span style="text-decoration: unde…<p><span style="vertical-align: baseline;">We recently announced the general availability (GA) of </span><strong style="vertical-align: baseline;">token-based </strong><a href="https://docs.cloud.google.com/bigquery/docs/control-genai-costs"><strong style="text…<h3><span style="vertical-align: baseline;">Quotas for managing costs</span></h3><p><span style="vertical-align: baseline;">Four daily quotas, managed through the Google Cloud Identity and Access Management (IAM) quota system, provide controls at the </span><strong style="vertical-align: baseline;">project</strong><span style="vertical-align: bas…@
@@ -1047,7 +1137,7 @@<p><code style="vertical-align: baseline;">AI.COUNT_TOKENS</code><span style="vertical-align: baseline;"> returns the number of tokens a model's tokenizer produces for a text string, without making a billable generation call. Run it over a column to measure your input data:&l…<div class="block-code"><dl> <dt>code_block</dt>-
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &quot;-- Count the input text tokens for each product review\r\nSELECT\r\n SUM(\r\n AI.COUNT_TOKENS(\r\n CONCAT(&#x27;Classify the sentiment of this review as positive, negative, or neutral: &#x27;, review…+
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &quot;-- Count the input text tokens for each product review\r\nSELECT\r\n SUM(\r\n AI.COUNT_TOKENS(\r\n CONCAT(&#x27;Classify the sentiment of this review as positive, negative, or neutral: &#x27;, review…</dl></div><div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The count is a lower bound on input tokens. It doesn't include output or thinking tokens, the instructions each AI function adds around your input, or non-text inputs such as images. Use it to:</spa…<ul>@
@@ -1074,7 +1164,7 @@<p><span style="vertical-align: baseline;">If a small sample run shows an average </span><strong style="vertical-align: baseline;">100:1 input-to-output token ratio</strong><span style="vertical-align: baseline;">, you can solve for a daily budget:</span><…<div class="block-code"><dl> <dt>code_block</dt>-
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &#x27;Cost: 0.125 × I + 0.75 × O = 100 (I, O in millions of tokens)\r\nRatio: I = 100 × O&#x27;), (&#x27;language&#x27;, &#x27;&#x27;), (&#x27;caption&#x27;, &lt;wagtail.rich_text.Ric…+
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &#x27;Cost: 0.125 × I + 0.75 × O = 100 (I, O in millions of tokens)\r\nRatio: I = 100 × O&#x27;), (&#x27;language&#x27;, &#x27;&#x27;), (&#x27;caption&#x27;, &lt;wagtail.rich_text.Ric…</dl></div><div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">Solving gives about </span><strong style="vertical-align: baseline;">754 million input tokens</strong><span style="vertical-align: baseline;"> and </span><strong style=…<h3><span style="vertical-align: baseline;">Best practices for managing AI cost</span></h3>@
@@ -2084,7 +2174,7 @@<p><span style="vertical-align: baseline;">Integrating cloud management into your agents no longer requires packaging Google Cloud CLI binaries, managing local execution runtimes, or maintaining dependencies inside agent container images. Because the Google Cloud CLI remote MCP server im…<div class="block-code"><dl> <dt>code_block</dt>-
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &#x27;{\r\n &quot;mcpServers&quot;: {\r\n &quot;google-cloud-cli&quot;: {\r\n &quot;uri&quot;: &quot;https://cloudcli.googleapis.com/mcp&quot;,\r\n ...\r\n }\r\n }\r\n}&…+
<dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &#x27;{\r\n &quot;mcpServers&quot;: {\r\n &quot;google-cloud-cli&quot;: {\r\n &quot;uri&quot;: &quot;https://cloudcli.googleapis.com/mcp&quot;,\r\n ...\r\n }\r\n }\r\n}&…</dl></div><div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">Your agent immediately gains access to execute gcloud and bq commands in a secure, network-isolated cloud sandbox.</span></p><p><span style="vertical-align: baseline;">Authentication is handled via keyless Agent Identity for hosted Google Cloud platforms, or standard OAuth 2.0 for external runtimes. For authentication options, see the </span><a href="https://docs.cloud.google.com/mcp/set-up-authentica…@
@@ -3267,444 +3357,4 @@<p><span style="vertical-align: baseline;">The sensible path runs in sequence. Establish the community practice, the way Orange did. Then introduce read-only agents that inform and suggest. Only once those are established across the community should you build agents that execute changes.…<p><span style="vertical-align: baseline;">How you build depends on who's building. For teams that want to deploy quickly with minimal code, the </span><a href="https://cloud.google.com/gemini-enterprise"><span style="text-decoration: underline; vertical-align: baseline;"&…<p><span style="vertical-align: baseline;">Cloud FinOps is moving beyond centralized reporting toward action that happens where the work does. The organizations getting there start with the culture, then use agents to carry it further than any one team could reach. </span></p>…-
<p><span style="vertical-align: baseline;">Orange's numbers came out of the community work. Building that foundation is the part worth copying first. When you're ready to extend it, </span><a href="https://cloud.google.com/consulting" rel="noopener" target="_blank"><span s…-
<div class="block-aside"><dl>-
<dt>aside_block</dt>-
<dd>&lt;ListValue: [StructValue([(&#x27;title&#x27;, &#x27;Get vital board insights with Google Cloud&#x27;), (&#x27;body&#x27;, &lt;wagtail.rich_text.RichText object at 0x7f421417cdd0&gt;), (&#x27;btn_text&#x27;, &#x27;Visit the hub&#x27…-
</dl></div>-
<div class="block-paragraph"><h3 data-block-key="hswvv"><b>‘Spellcheck for cybersecurity’ and beyond: How Google monitors AI threats and advances AI defenses</b></h3><p data-block-key="61crd"><i>By Sandra Joyce, VP, Google Threat Intelligence</i></p…-
<div class="block-paragraph_with_image"><div class="article-module h-c-page">-
<div class="h-c-grid uni-paragraph-wrap">-
<div class="uni-paragraph-
h-c-grid__col h-c-grid__col--8 h-c-grid__col-m--6 h-c-grid__col-l--6-
h-c-grid__col--offset-2 h-c-grid__col-m--offset-3 h-c-grid__col-l--offset-3">-
-
-
-
-
-
-
-
-
-
-
<figure class="article-image--wrap-small-
-
"-
>-
-
-
-
-
<img-
src="https://storage.googleapis.com/gweb-cloudblog-publish/images/2022_S_Joyce_Headshot.max-1000x1000.jpg"-
-
alt="Sandra Joyce">-
-
</a>-
-
<figcaption class="article-image__caption "><p data-block-key="nj7d4">Sandra Joyce, VP, Google Threat Intelligence</p></figcaption>-
-
</figure>-
-
-
-
-
-
-
-
<p data-block-key="0jyqm">Anyone operating in security knows that speculation is a major liability during periods of technological disruption. While there is plenty of hype and understandable concern around how threats might use and target AI, a CISO’s AI security strategy has to be anch…-
</div>-
</div>-
</div>-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">When we strip away the noise and look at the telemetry, the real threat landscape boils down to three structural shifts that CISOs must address:</span></p>-
<ol>-
<li aria-level="1" style="list-style-type: decimal; vertical-align: baseline;">-
<p role="presentation"><span style="vertical-align: baseline;">AI is reshaping how software is built. </span></p>-
</li>-
<li aria-level="1" style="list-style-type: decimal; vertical-align: baseline;">-
<p role="presentation"><span style="vertical-align: baseline;">AI is expanding the attack surface.</span></p>-
</li>-
<li aria-level="1" style="list-style-type: decimal; vertical-align: baseline;">-
<p role="presentation"><span style="vertical-align: baseline;">AI is enhancing threat capabilities. </span></p>-
</li>-
</ol>-
<p><span style="vertical-align: baseline;">Today, we’re sharing details on Google’s visibility into these three challenges, and our approach for solving them.</span></p>-
<p><strong style="vertical-align: baseline;">Building securely in the AI era </strong></p>-
<p><span style="vertical-align: baseline;">AI has fundamentally altered software development velocity. Across the industry, autonomous agents and AI workflows now push code into production at unprecedented speed. This creates exciting opportunities for innovation, yet CISOs are faced wit…-
<p><span style="vertical-align: baseline;">We’re seeing threat actors turn our greatest engineering shortcut against us by contaminating upstream packages that AI assistants are trained to suggest and trust. GTIG believes that malicious contamination of AI-assisted coding practices has b…-
<div class="block-pull_quote"><div class="uni-pull-quote h-c-page">-
<section class="h-c-grid">-
<div class="uni-pull-quote__wrapper h-c-grid__col h-c-grid__col--8 h-c-grid__col-m--6 h-c-grid__col-l--6-
h-c-grid__col--offset-2 h-c-grid__col-m--offset-3 h-c-grid__col-l--offset-3">-
<div class="uni-pull-quote__inner-wrapper h-c-copy h-c-copy">-
<q class="uni-pull-quote__text">The solution to a machine-speed threat landscape isn&#x27;t slowing developers down — it’s building security natively into the AI pipeline. Part of this process involves in-editor guardrails for developers that create a real-time &#x27;spellcheck…-
-
-
</div>-
</div>-
</section>-
</div>-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">We’re also monitoring adversaries targeting agents. The financially-motivated threat actor TeamPCP (UNC6780) has implemented more than half a dozen methods to exploit AI tools and open-source software …-
<p><span style="vertical-align: baseline;">The solution to a machine-speed threat landscape isn't slowing developers down — it’s building security natively into the AI pipeline. Part of this process involves in-editor guardrails for developers that create a real-time “spellcheck for cybe…-
<p><span style="vertical-align: baseline;">Just as word processors underline typos without forcing the writer to stop, security controls must sit natively inside the developer’s editor and agentic workflows, instantly flagging poisoned packages, toxic prompts, and misconfigured toolkits.…-
<p><span style="vertical-align: baseline;">Crucially, this can’t stop at the editor. Traditional security suffers from context blindness: Code editors can’t see cloud configurations, delivery pipelines miss runtime exposure, and production teams can’t easily patch root-cause blueprints. …-
<p><span style="vertical-align: baseline;">Bridging this gap requires an integrated code-to-cloud approach — the exact design principle behind platforms like </span><a href="https://www.wiz.io/platform/wiz-code" rel="noopener" target="_blank"><span style="text-decoration: …-
<p><span style="vertical-align: baseline;">When organizations think about AI-driven code analysis, the default assumption is to pick one frontier model and point it at their repository. However, our research and telemetry show that single-model security creates a dangerous monoculture: N…-
<div class="block-pull_quote"><div class="uni-pull-quote h-c-page">-
<section class="h-c-grid">-
<div class="uni-pull-quote__wrapper h-c-grid__col h-c-grid__col--8 h-c-grid__col-m--6 h-c-grid__col-l--6-
h-c-grid__col--offset-2 h-c-grid__col-m--offset-3 h-c-grid__col-l--offset-3">-
<div class="uni-pull-quote__inner-wrapper h-c-copy h-c-copy">-
<q class="uni-pull-quote__text">To secure this expanding attack surface, CISOs should avoid the trap of managing AI through disconnected silos... The future of cloud and AI defense needs to be built on a unified and dynamic graph that connects your code, your models, your data lineage,…-
-
-
</div>-
</div>-
</section>-
</div>-
-
</div>-
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">To solve this, Google takes a </span><a href="https://cloud.google.com/blog/products/identity-security/cloud-ciso-perspectives-next-26-why-we-re-multicloud-and-multi-ai"><span style="tex…-
<p><strong style="vertical-align: baseline;">Securing AI </strong></p>-
<p><span style="vertical-align: baseline;">Securing the development lifecycle is only half the battle. We also need to prevent adversaries from exploiting AI attack surfaces and weaponizing over-privileged agents. Threat actors are targeting AI workloads with techniques that include: <…-
<ul>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">LLMJacking</strong><span style="vertical-align: baseline;">: Cybercriminals and state-sponsored groups target GPU access to support running their AI models and agentic workflows. In one notable intrusion Mandian…-
</li>-
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">-
<p role="presentation"><strong style="vertical-align: baseline;">Targeting of AI data and access</strong><span style="vertical-align: baseline;">: Cybercriminals now recognize that your custom prompts, agent instructions, and fine-tuned models represent high-value crown jewel…-
</li>-
</ul>-
<p><span style="vertical-align: baseline;">To secure this expanding attack surface, CISOs should avoid the trap of managing AI through disconnected silos. Don’t treat agent access policies, model inventories (AI-BOMs) and shadow AI as separate challenges because these risks are deeply co…-
<p><span style="vertical-align: baseline;">Pioneered by the </span><a href="https://www.wiz.io/lp/wiz-security-graph" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Wiz Security Graph</span></a><span sty…-
<p><strong style="vertical-align: baseline;">Defending against AI threats</strong></p>-
<p><span style="vertical-align: baseline;">Threat actors are rapidly moving beyond simple prompt generation toward fully-automated, multi-agent attack pipelines.</span></p></div>-
<div class="block-video">-
-
-
-
<div class="article-module article-video ">-
<figure>-
<a class="h-c-video h-c-video--marquee"Diff display stops at 400 lines. The line counts above are from the whole diff. 115 lines shown here cut at 300 characters. The raw artifact at this commit is linked above.