广告
加载中

谷歌开放Gemini搜索第三方接入 推出AI开发工具与XR眼镜

亿邦AI 2026-10-29 09:29
亿邦AI 2026/10/29 09:29

邦小白快读

EN
全文速览

总1:峰会主线是谷歌把搜索、Gemini、安卓开发与XR硬件连成开放生态。

1. 普通用户可通过搜索或Gemini绑定第三方服务,完成设计宣传单、创建播放列表等操作,Gemini端已有50多款互联应用。

2. 开发者可在Android Studio内运行40多款外部代码智能体,CLI可让智能体渲染Compose预览,官方称UI任务速度提升77%、token消耗最多降三分之二。

总2:可感知的新硬件与体验。

1. XREAL AURA智能眼镜重不足95克,70度视场角,售价1099美元,美国、加拿大、韩国可预订,11月中旬发货。

2. 眼镜可同屏放5款安卓应用、USB-C投屏笔记本,内置Gemini可识别环境并整理窗口、游戏提示、高亮实体物品。

总1:品牌可借谷歌互联应用平台进入搜索与Gemini双入口。

1. 企业可自主提交互联应用,选择搜索端、Gemini端或双端上线,并追踪流量、延迟、错误率等运营数据。

2. 开发者保留认证logo、账号验证等品牌标识控制权,有利于品牌露出与信任建设。

3. Gemini端已有Adobe、益博睿、Localiza、Peloton等50多款应用,谷歌自有Workspace、Photos、YouTube Music、Google Home也在矩阵内。

总2:营销与产品研发启示。

1. 品牌可把服务嵌入AI助手和搜索场景,如Canva设计、YouTube Music播放列表,缩短用户从需求到服务的路径。

2. XR眼镜提供新品牌触点,内置Gemini可做环境识别、窗口整理和游戏提示;安卓智能体转UI接口支持应用内组件展示可交互界面。

3. 现阶段仅美国开发者候补预览,品牌应关注双端上线节奏与运营数据,不宜把大范围投放当作既定前提。

总1:平台开放带来触达与转化新机会。

1. 谷歌互联应用平台面向希望经搜索或Gemini触达用户的企业和开发者开放,可提交应用并监测流量、延迟、错误率。

2. Gemini端已上线50多款互联应用,合作方包括Adobe、益博睿、Localiza、Peloton;搜索端已接入Instacart、Canva、YouTube Music、雅虎梦幻足球,Calm将双端上线。

3. 卖家可学习把服务封装为可调用应用,通过云托管MCP服务器或AppFunctions API接入安卓端MCP。

总2:机会与风险提示。

1. 机会在于AI助手和搜索成为服务入口,用户可安全绑定并直接调用第三方服务完成特定操作。

2. 风险是公开预览版仅面向加入候补名单的美国开发者,大范围上线时间未公布,投入需控制节奏。

3. 可关注双端上线选择、品牌标识控制和运营指标,作为合作方式与效果评估依据。

总1:XR硬件披露了明确的产品设计与生产参数。

1. XREAL AURA是首款运行Android XR的有线智能眼镜,由谷歌联合硬件厂商XREAL发布,重不足95克。

2. 采用70度视场角透明可调光镜片,计算单元与电池集成在外置触控板模块,搭载高通Snapdragon Reality Elite芯片。

3. 售价1099美元,已在美国、加拿大、韩国开启预订,11月中旬发货,可作为硬件制造与供应链机会参考。

总2:数字化与开发生态启示。

1. Jetpack XR SDK、Android Studio内XR模拟器、面向Unity、Godot、Unreal的Engine Hub均推出稳定版,降低XR内容制作门槛。

2. ML Kit新增Gemini Nano 4驱动的Agentic API,AppFunctions API作为统一工具调用层,利于端侧AI功能集成。

3. 眼镜可同屏放5款安卓应用、USB-C投屏笔记本,内置Gemini识别真实环境与虚拟屏幕,为产品功能设计提供方向。

总1:开发者工具围绕智能体协作和效率提升升级。

1. Android Studio Rabbit 2支持在IDE内运行40多款外部代码智能体,包括Claude Agent、OpenAI Codex、谷歌Antigravity,可访问项目结构、符号解析、模拟器和Logcat。

2. CLI工具成为Claude Code、OpenAI Codex、谷歌Antigravity官方插件,智能体可在终端渲染Compose预览;谷歌内测称UI任务速度提升77%、token最多降三分之二。

3. LeakCanary内置CLI与分析器;基于Claude Opus 5.5的A/B测试称内存泄漏排查修复速度提升近一倍、token降约25%。

总2:客户痛点与解决方案。

1. 针对大模型最新API训练不足,谷歌上线30多项安卓技能模块,覆盖Play政策检查、权限安全、折叠屏及Googlebooks布局适配。

2. Change Review可分析待提交代码并生成漏洞报告;迁移助手可把旧XML布局转Jetpack Compose;Lightbuild支持人机共同读写。

3. 端侧有Agentic API、AppFunctions API和智能体转UI接口,服务商可围绕安全过滤、工具调用和可交互界面做方案。

总1:谷歌正用开放平台把搜索、Gemini和安卓开发者生态连接起来。

1. 互联应用平台降低企业接入门槛,开发者通过云托管MCP服务器或AppFunctions API接入安卓端MCP,在互联控制台完成注册、测试与提交,可选择搜索端、Gemini端或双端上线。

2. 平台提供流量、延迟、错误率等运营数据,并保留认证logo、账号验证等品牌控制权;Gemini端已有50多款应用,搜索端接入Instacart、Canva、YouTube Music、雅虎梦幻足球等。

总2:平台运营与风向规避。

1. 安卓开发工具开放40多款外部代码智能体,CLI成为多款智能体官方插件,基准排行榜从功能、耗时、成本评估大模型,体现平台招商与生态运营思路。

2. XR开发环境转正,Jetpack XR SDK、XR模拟器和Engine Hub稳定版上线,并联合XREAL推出AURA眼镜。

3. 风险是互联应用平台公开预览仅限美国候补开发者,大范围上线未公布;平台需关注接入审核、品牌控制、安全过滤和运营指标。

总1:文章显示AI助手与搜索正演变为第三方服务分发和互操作平台。

1. 谷歌互联应用平台支持企业经搜索或Gemini接入,用户可安全绑定并直接调用第三方服务;Gemini端已有50多款应用,覆盖Adobe、益博睿、Localiza、Peloton及谷歌自有服务。

2. 接入依赖云托管MCP服务器或AppFunctions API,平台保留认证logo和账号验证控制,并提供流量、延迟、错误率数据,值得研究平台治理与接口标准化。

总2:新问题与商业模式。

1. Android Studio可运行40多款外部代码智能体,CLI提升UI任务速度77%、token最多降三分之二;基准排行榜从功能、耗时、成本评价模型,提出智能体效能度量问题。

2. 端侧Agentic API、AppFunctions API和智能体转UI接口,使智能体从文本走向可交互界面,涉及安全过滤、权限与工具调用边界。

3. XR眼镜售价1099美元、重不足95克、70度视场角,Android XR生态转正,可作为硬件商业模式与人机交互研究样本。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

Summary 1: The keynote's throughline was Google linking Search, Gemini, Android development, and XR hardware into an open ecosystem.

1. Ordinary users can bind third-party services through Search or Gemini to complete tasks such as designing flyers and creating playlists; Gemini already has more than 50 connected apps.

2. Developers can run more than 40 external code agents inside Android Studio; the CLI lets agents render Compose previews; Google says UI tasks are 77% faster and token consumption is reduced by up to two-thirds.

Summary 2: Tangible new hardware and experiences.

1. XREAL AURA smart glasses weigh under 95 grams, have a 70-degree field of view, cost $1,099, are available for preorder in the U.S., Canada, and South Korea, and ship in mid-November.

2. The glasses can hold five Android apps on one screen and mirror a laptop via USB-C; the built-in Gemini can recognize the environment and organize windows, offer game hints, and highlight physical objects.

Summary 1: Brands can enter both Search and Gemini entry points through Google's connected-app platform.

1. Companies can submit connected apps themselves, choose Search-only, Gemini-only, or dual-platform launch, and track operational data such as traffic, latency, and error rates.

2. Developers retain control over brand identifiers such as verified logos and account verification, which helps with brand exposure and trust building.

3. Gemini already has more than 50 connected apps, including Adobe, Experian, Localiza, and Peloton; Google's own Workspace, Photos, YouTube Music, and Google Home are also in the matrix.

Summary 2: Implications for marketing and product development.

1. Brands can embed services into AI assistant and Search scenarios, such as Canva design and YouTube Music playlists, shortening the path from user need to service.

2. XR glasses provide a new brand touchpoint; built-in Gemini can perform environment recognition, window organization, and game hints; Android's agent-to-UI interface supports in-app components that display interactive interfaces.

3. At this stage, only U.S. developers can join the waitlist preview, so brands should monitor the dual-platform launch timeline and operational data and should not treat broad rollout as a given.

Summary 1: Platform openness creates new opportunities for reach and conversion.

1. Google's connected-app platform is open to businesses and developers seeking to reach users through Search or Gemini; they can submit apps and monitor traffic, latency, and error rates.

2. More than 50 connected apps are live on Gemini, with partners including Adobe, Experian, Localiza, and Peloton; Search has integrated Instacart, Canva, YouTube Music, and Yahoo Fantasy Football, and Calm will launch on both platforms.

3. Sellers can learn to package services as callable apps and connect to Android-side MCP through cloud-hosted MCP servers or the AppFunctions API.

Summary 2: Opportunity and risk.

1. The opportunity is that AI assistants and Search become service entry points, allowing users to securely bind and directly invoke third-party services for specific tasks.

2. The risk is that the public preview is limited to U.S. developers on the waitlist, with no broad launch date announced, so investment should be paced.

3. Sellers should watch dual-platform launch choices, brand-identifier control, and operational metrics as the basis for partnership approaches and performance evaluation.

Summary 1: The XR hardware disclosure provides clear product-design and production parameters.

1. XREAL AURA is the first wired smart glasses running Android XR, released by Google with hardware maker XREAL, and weighs under 95 grams.

2. It uses transparent, dimmable lenses with a 70-degree field of view; the computing unit and battery are integrated into an external touchpad module, and it runs on Qualcomm's Snapdragon Reality Elite chip.

3. Priced at $1,099, it is already available for preorder in the U.S., Canada, and South Korea and ships in mid-November, offering a reference point for hardware manufacturing and supply-chain opportunities.

Summary 2: Implications for digitalization and the developer ecosystem.

1. The Jetpack XR SDK, XR emulator in Android Studio, and Engine Hub for Unity, Godot, and Unreal have all reached stable release, lowering the barrier to XR content creation.

2. ML Kit adds an Agentic API powered by Gemini Nano 4, and the AppFunctions API serves as a unified tool-calling layer, helping with on-device AI feature integration.

3. The glasses can hold five Android apps on one screen and mirror a laptop via USB-C; built-in Gemini recognizes the real environment and virtual screens, offering direction for product feature design.

Summary 1: Developer tools are upgraded around agent collaboration and efficiency gains.

1. Android Studio Rabbit 2 supports running more than 40 external code agents inside the IDE, including Claude Agent, OpenAI Codex, and Google Antigravity; these can access project structure, symbol resolution, emulators, and Logcat.

2. The CLI tool becomes an official plugin for Claude Code, OpenAI Codex, and Google Antigravity, allowing agents to render Compose previews in the terminal; Google's internal testing claims UI tasks are 77% faster and tokens are reduced by up to two-thirds.

3. LeakCanary includes a built-in CLI and analyzer; an A/B test based on Claude Opus 5.5 claims memory-leak diagnosis and fixes are nearly twice as fast, with token use down about 25%.

Summary 2: Customer pain points and solutions.

1. Addressing insufficient training on the latest LLM APIs, Google launched more than 30 Android skill modules covering Play policy checks, permission security, foldable-screen and Googlebooks layout adaptation.

2. Change Review can analyze code pending submission and generate vulnerability reports; the migration assistant can convert legacy XML layouts to Jetpack Compose; Lightbuild supports human-machine co-reading and writing.

3. On-device offerings include the Agentic API, AppFunctions API, and agent-to-UI interface, giving service providers room to build solutions around security filtering, tool calling, and interactive interfaces.

Summary 1: Google is using an open platform to connect Search, Gemini, and the Android developer ecosystem.

1. The connected-app platform lowers the barrier for enterprise access; developers connect to Android-side MCP through cloud-hosted MCP servers or the AppFunctions API, complete registration, testing, and submission in the connected console, and can choose a Search-only, Gemini-only, or dual-platform launch.

2. The platform provides operational data such as traffic, latency, and error rates and preserves brand control over verified logos and account verification; Gemini already has more than 50 apps, while Search has integrated Instacart, Canva, YouTube Music, Yahoo Fantasy Football, and others.

Summary 2: Platform operations and risk avoidance.

1. Android development tools open more than 40 external code agents, the CLI becomes an official plugin for multiple agents, and a benchmark leaderboard evaluates large models by capability, time, and cost, reflecting the platform's approach to attracting partners and operating the ecosystem.

2. The XR development environment has moved to general availability, with stable releases of the Jetpack XR SDK, XR emulator, and Engine Hub, and Google has partnered with XREAL to launch AURA glasses.

3. The risk is that the connected-app platform's public preview is limited to U.S. waitlisted developers, with no broad launch announced; the platform needs to focus on access review, brand control, security filtering, and operational metrics.

Summary 1: The article shows AI assistants and Search are evolving into distribution and interoperability platforms for third-party services.

1. Google's connected-app platform supports enterprise access via Search or Gemini, allowing users to securely bind and directly invoke third-party services; Gemini already has more than 50 apps, covering Adobe, Experian, Localiza, Peloton, and Google's own services.

2. Access relies on cloud-hosted MCP servers or the AppFunctions API; the platform retains control over verified logos and account verification and provides traffic, latency, and error-rate data, making platform governance and interface standardization worth studying.

Summary 2: New questions and business models.

1. Android Studio can run more than 40 external code agents; the CLI improves UI task speed by 77% and reduces tokens by up to two-thirds; a benchmark leaderboard evaluates models by capability, time, and cost, raising questions about measuring agent effectiveness.

2. On-device Agentic API, AppFunctions API, and agent-to-UI interfaces move agents from text to interactive interfaces, involving security filtering, permissions, and boundaries of tool calling.

3. XR glasses priced at $1,099, weighing under 95 grams, with a 70-degree field of view, and the Android XR ecosystem reaching general availability can serve as research samples for hardware business models and human-computer interaction.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

2026年10月28日,谷歌在安卓开发者峰会上发布多项产品更新,覆盖搜索与Gemini互联生态、安卓开发工具、端侧AI接口及XR硬件多个领域。

当天谷歌正式推出互联应用平台,降低企业将自有应用接入谷歌搜索与Gemini AI体验的门槛。今年早些时候,互联应用功能已在Gemini与谷歌搜索上线,支持用户安全绑定并直接调用第三方服务完成特定操作,包括接入Canva设计宣传单、接入YouTube Music创建播放列表等。目前ChatGPT、Meta Muse等面向消费者的AI服务均已推出类似的第三方服务集成功能。新平台面向所有希望通过搜索或Gemini触达用户的企业与开发者开放,支持其自主提交互联应用并监测运营表现。目前平台预览版已向美国开发者开放beta测试,接入流程为开发者通过云托管的模型上下文协议服务器,或是结合AppFunctions API接入安卓端模型上下文协议,在谷歌互联控制台完成应用注册、测试与提交流程,可自主选择应用落地在搜索端、Gemini端或是双端同步上线。应用正式上线后,开发者可追踪流量、延迟、错误率等运营数据,保留认证logo、账号验证等品牌标识控制权。

现阶段Gemini端已上线超过50款互联应用,合作方包括Adobe、益博睿、Localiza、Peloton等,覆盖谷歌自有Workspace、Photos、YouTube Music、Google Home等全量服务矩阵。这些应用将在未来几周内逐步登陆谷歌搜索。当前谷歌搜索端已接入Instacart、Canva、YouTube Music、雅虎梦幻足球等应用,冥想应用Calm作为新合作方,其服务将很快在Gemini与搜索双端上线。互联应用平台公开预览版目前仅面向加入候补名单的美国开发者开放,大范围上线时间尚未公布。

安卓开发工具升级是本次峰会的核心内容。随Android Studio Rabbit 2版本同步上线的自带智能体功能,打破了此前开发者仅能使用谷歌自有AI工具的限制。基于开放的智能体客户端协议,开发者可在集成开发环境内直接运行超过40款外部代码智能体,包括Anthropic的Claude Agent、OpenAI的Codex、谷歌Antigravity,这些智能体可直接访问项目结构、符号解析、模拟器、Logcat系统等开发工具。同步上线的2.0版本安卓基准排行榜,将从功能实现、耗时、成本三个维度,评估大模型在多日开发项目中的实际表现。新增的代码验证工具Change Review可分析待提交代码改动,生成漏洞报告,逐步讲解修改细节。Android Studio Canary预览版新增迁移助手,可自动将旧版XML布局转换为Jetpack Compose格式。谷歌同时在测试基于YAML schema的声明式构建系统Lightbuild,支持人类开发者与AI智能体共同读写编辑。

针对在终端或CI流水线运行智能体的开发者,谷歌扩展了安卓CLI命令行工具,目前该工具已成为Claude Code、OpenAI Codex、谷歌Antigravity的官方智能体插件。新增命令支持AI智能体在终端渲染Compose预览,无需启动模拟器即可直观检查UI改动,谷歌内部测试数据显示,该功能可让智能体完成UI任务的速度提升77%,token消耗最多降低三分之二。CLI工具还支持接入谷歌实验室的流传输实体测试设备,方便智能体排查设备专属漏洞。调试与性能分析板块新增多项功能,开源内存泄漏检测库LeakCanary已内置到CLI与Android Studio分析器中,将堆转储评估环节放在电脑端运行,避免测试过程拖慢应用运行速度。谷歌使用Claude Opus 5.5模型开展的A/B测试显示,该CLI工具可将内存泄漏的排查修复速度提升近一倍,token消耗降低约25%。其余命令行工具覆盖多项开发场景,包括通过安卓性能分析器捕获系统追踪记录、从Firebase Crashlytics拉取崩溃报告、通过R8编译器检查代码优化情况。针对大模型对最新API训练不足的问题,谷歌上线超过30项安卓技能模块,向智能体同步最新开发文档,覆盖应用上架前谷歌Play政策检查、权限安全问题处理、折叠屏及Googlebooks笔记本设备布局适配等场景。

面向端侧AI功能开发,谷歌在ML Kit中新增由Gemini Nano 4驱动的Alpha版Agentic API,可自动执行函数、保存聊天记录、管理编排循环、提供安全过滤。谷歌还基于模型上下文协议推出beta版AppFunctions API,作为统一工具调用层支持应用向本地或全系统智能体共享功能。配套上线的还有面向Jetpack Compose的Alpha版智能体转UI接口,智能体无需输出纯文本块,可直接基于应用内组件展示可交互用户界面。

系统适配层面,Jetpack Compose 1.13与AppState模型已完成大尺寸屏幕适配优化,XR开发环境结束测试阶段,Jetpack XR SDK、Android Studio内的XR模拟器、面向Unity、Godot、Unreal引擎的Engine Hub均推出正式稳定版本。谷歌联合硬件厂商XREAL发布XREAL AURA智能眼镜,这是首款运行Android XR系统的有线智能眼镜,重量不足95克,配备70度视场角的透明可调光镜片,计算单元与电池集成在外置的触控板模块中,搭载高通全新Snapdragon Reality Elite芯片。该眼镜售价1099美元,目前已在美国、加拿大、韩国开启预订,11月中旬启动发货。用户可在视野内同时放置最多5款安卓应用,也可通过USB-C线缆投屏笔记本屏幕。眼镜内置Gemini,可通过镜头识别真实环境与虚拟屏幕,实现窗口整理、游戏攻略提示、房间内实体物品视觉高亮等功能。

本文首发于 亿邦动力 官方网站

文章来源:亿邦动力

广告
微信
朋友圈

FAQ回顾

企业如何将自有应用接入谷歌搜索和Gemini?

开发者可通过云托管的模型上下文协议服务器,或结合AppFunctions API接入安卓端模型上下文协议,在谷歌互联控制台完成注册、测试和提交,并选择应用上线搜索端、Gemini端或双端。上线后可追踪流量、延迟、错误率等数据,并保留认证logo、账号验证等品牌控制权。

Gemini和谷歌搜索目前接入了哪些第三方互联应用?

Gemini端已上线超过50款互联应用,合作方包括Adobe、益博睿、Localiza、Peloton,并覆盖Workspace、Photos、YouTube Music、Google Home等服务;谷歌搜索端已接入Instacart、Canva、YouTube Music、雅虎梦幻足球,Calm将很快在双端上线。

Android Studio的自带智能体功能支持哪些外部AI代码智能体?

Android Studio Rabbit 2的自带智能体功能基于开放智能体客户端协议,允许开发者在IDE内运行超过40款外部代码智能体,包括Anthropic Claude Agent、OpenAI Codex和谷歌Antigravity;这些智能体可访问项目结构、符号解析、模拟器、Logcat等开发工具。

XREAL AURA智能眼镜是什么产品,有哪些核心配置和功能?

XREAL AURA是首款运行Android XR系统的有线智能眼镜,重不足95克,配70度视场角透明可调光镜片,计算单元和电池外置于触控板,搭载高通Snapdragon Reality Elite芯片,售价1099美元,11月中旬发货;可在视野内放置最多5款安卓应用,支持USB-C投屏,并内置Gemini进行环境识别和窗口整理。

谷歌为端侧AI开发提供了哪些API工具?

谷歌在ML Kit中新增由Gemini Nano 4驱动的Alpha版Agentic API,可自动执行函数、保存聊天记录、管理编排循环并提供安全过滤;同时推出beta版AppFunctions API作为统一工具调用层,以及面向Jetpack Compose的Alpha版智能体转UI接口。

这么好看,分享一下?

朋友圈 分享

APP内打开

赞 +1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0