扩展 Grounding 提供商选择：与 Parallel Web Systems 达成合作
随着生成式 AI 从简单的聊天界面演进为执行复杂任务的智能体，例如自动化 KYC（了解你的客户）核查、商品目录数据增强、实时新闻分析以及企业尽职调查，对事实准确性的需求前所未有地高涨。
为了让开发者能够更灵活地将 Gemini 模型基于可验证的实时信息进行 grounding，今天我们很高兴地宣布，Parallel Web Systems 现已成为 Gemini Enterprise Agent Platform 上原生集成的网络 grounding 提供商。
使用 Parallel Web Search 进行 Grounding可将 Gemini 模型锚定在高质量、实时的网络结果之上。通过提供对原始来源的精确引用，它确保你的智能体基于准确且可验证的信息采取行动。该能力现已在 Agent Platform 全面可用：可在 Gemini API 中调用，可在 Agent Studio 中选择，并可通过 Google Cloud Marketplace 订阅，用量计入你现有的 Google Cloud 账单。
“企业希望在开放平台上构建 AI 智能体，借助最优秀的工具创造真正的价值。通过将 Parallel 作为我们 Gemini Enterprise Agent Platform 上的一个 grounding 选项加入，客户现在可以使用他们偏好的实时网络数据流，让 Gemini 变得更智能。这帮助他们顺畅地将智能体从原型推进到部署，而且全程都在他们现有的 Google Cloud 环境中完成。”——Matt Renner，Google Cloud 总裁兼首席营收官。
“AI 智能体很快对网络的使用将远超人类以往的任何时候，它们需要专为其运作方式打造的搜索基础设施。我们构建 Parallel 正是为了提供这样的基础设施，而与 Google Cloud 的这次合作将它原生地带到了企业构建生产级智能体的地方。”——Parag Agrawal，Parallel 创始人兼 CEO。
关于 Parallel Web Systems
Parallel Web Systems 构建专为智能体打造的搜索基础设施。Parallel 的 Search API 由专为智能体工作负载构建的专有网络索引驱动，提供结构化的、针对 LLM 优化的结果。通过与金融服务、法律和技术领域的企业合作，Parallel 助力大规模地驱动生产级自主智能体工作流。
将 Gemini 与 Parallel Web Search 连接起来
在 Agent Platform 上通过 Parallel Web Search 进行接地（Grounding），可将 Gemini 模型连接到由 Parallel Web Systems 的搜索 API 提供的公开网络数据。
该服务将 Parallel Web Systems 的网络索引与 Gemini 的原生能力相结合，能够理解并拆解复杂的用户提示词，从最相关的搜索结果中提炼洞见，并生成带有精确引用标注的高准确度答案。由于 Parallel Web Search 运行在 Google Cloud 上，你可以受益于直接在现有云环境中实现的精简、安全的集成，并且针对敏感工作负载还提供零数据留存（ZDR）选项。
以扩展的灵活性解锁复杂架构
将 Parallel 加入我们的生态系统，为客户带来更广泛的许可和架构灵活性。Grounding with Parallel Web Search 的一个关键差异化优势，在于它以开放的方式支持新的用例。对开发者而言，这意味着可以自由地大规模执行程序化调用、提取并缓存网络数据以丰富内部数据集，并使用其他 LLM 对 Grounding with Parallel Web Search 的结果进行后处理。
这种灵活性支持复杂架构，让企业能够放心地在生产环境中运行智能体，并将 Gemini 的网络接地能力集成到其现有技术栈中。
这种扩展的灵活性使开发者能够构建高度专业化的智能体工作流，例如：
- 目录与数据库增强： 提取并永久存储可验证的网页数据，以填补庞大内部系统中的空白。例如，管理大量产品库存的组织可以提取并永久存储网页数据，以持续丰富目录属性，或利用搜索数据自动补充和更新庞大的供应商与联系人数据库。
- 自主智能体： 构建能够持续导航系统、检索实时信息并执行复杂程序化工作流的智能体。例如，具有严格监管要求的组织可以构建智能体，自动将内部文档与公共领域进行交叉比对，或针对实时网页数据以程序化方式运行全球合规与风险评估检查，无需人工干预。
- 多智能体编排： 构建复杂的多智能体系统，由中央编排器在不同模型之间路由任务。由于 Grounding with Parallel Web Search 允许开发者将搜索结果传递给其他 LLM，他们可以创建助手，将上下文和查询无缝传递给由不同 LLM 驱动的、面向特定任务的各种子智能体，从而提供高度综合的洞察。
立即开始
将 Grounding 与 Parallel Web Search 集成非常简单，并且直接内置于你的 Agent Platform 工作流中。要开始使用，请通过 Google Cloud Marketplace 订阅 Parallel 服务，接受服务条款，并查看定价。购买金额会计入你符合条件的 Marketplace 权益，并合并计入你的 Google Cloud 账单。
订阅后，前往 Agent Studio。在导航菜单中，点击 + New 并选择 Chat。展开 Model settings 面板，点击 Grounding with Partners 开关，打开 Customize partner grounding 菜单。选择 Parallel Web Search 作为你的 grounding 来源，应用配置，你的 Gemini 模型将立即准备好生成基于可验证、实时网络信息的回答。
Expanding Grounding Provider Choice: Partnership with Parallel Web Systems
As generative AI evolves from simple chat interfaces to agents executing complex tasks like automated Know Your Customer (KYC) checks, catalog data enrichment, real-time news analysis, and corporate due diligence, the need for factual accuracy has never been higher.
To give developers more flexibility in how they ground Gemini models to verifiable, real-time information, today, we are excited to announce that Parallel Web Systems is now a natively integrated web grounding provider on Gemini Enterprise Agent Platform.
Grounding with Parallel Web Search anchors Gemini models in high-quality, real-time web results. By providing exact citations to original sources, it ensures your agents act on information that's accurate and verifiable. This capability is now available across Agent Platform: callable in the Gemini API, selectable in Agent Studio, and available to subscribe through Google Cloud Marketplace, with usage metered on your existing Google Cloud invoice.
"Enterprises want to build AI agents on an open platform with the best tools to deliver real value. By adding Parallel as a grounding option on our Gemini Enterprise Agent Platform, customers can now use their preferred live-web data stream to make Gemini smarter. This helps them smoothly take agents from prototype to deployment, all within their existing Google Cloud environment."–Matt Renner, President and Chief Revenue Officer, Google Cloud.
"AI agents will soon use the web far more than humans ever have, and need search infrastructure purpose-built for how they operate. We built Parallel to provide that infrastructure, and this partnership with Google Cloud brings it natively to where enterprises are building production agents."–Parag Agrawal, Founder and CEO, Parallel.
About Parallel Web Systems
Parallel Web Systems builds search infrastructure purpose-built for agents. Parallel’s Search API delivers structured, LLM-optimized results powered by a proprietary web index built specifically for agentic workloads. By working with enterprises across financial services, legal, and technology, Parallel helps power production-grade autonomous agent workflows at scale.
Connecting Gemini with Parallel Web Search
Grounding with Parallel Web Search on Agent Platform connects Gemini models to public web data provided by Parallel Web Systems' search API. This service combines Parallel Web System’s web index with Gemini's native ability to understand and decompose complex user prompts, distill insights from the most relevant search results, and generate highly accurate answers complete with precise citation annotations. Because Parallel Web Search runs on Google Cloud, you benefit from a streamlined, secure integration directly within your existing cloud environment, with an option for zero data retention (ZDR) available for sensitive workloads.
Unlocking complex architectures with expanded flexibility
Adding Parallel to our ecosystem empowers customers with expanded licensing and architectural flexibility. A key differentiator of Grounding with Parallel Web Search is its expansive approach to new use cases. For developers, this means the freedom to execute programmatic calls at scale, extract and cache web data to enrich internal datasets, and post-process Grounding with Parallel Web Search results using other LLMs. This flexibility supports complex architectures, allowing enterprises to confidently run agents in production and integrate Gemini's web-grounding capabilities into their existing tech stacks.
This expanded flexibility enables developers to build highly specialized agentic workflows, such as:
- Catalog & database enrichment: Extract and permanently store verifiable web data to fill gaps in massive internal systems. For example, organizations managing large product inventories can extract and permanently store web data to continuously enrich catalog attributes, or leverage search data to automatically supplement and update vast vendor and contact databases.
- Autonomous Agents: Build agents that continuously navigate systems, retrieve real-time information, and execute complex programmatic workflows. For instance, organizations with strict regulatory requirements can build agents that automatically cross-reference internal documents against public domains, or programmatically run global compliance and risk-assessment checks against live web data without human intervention.
- Multi-Agent Orchestration: Build sophisticated multi-agent systems where a central orchestrator routes tasks across different models. Because Grounding with Parallel Web Search allows developers to pass search results to other LLMs, they can create assistants that seamlessly pass context and queries to diverse, task-specific sub-agents powered by different LLMs to deliver highly synthesized insights.
Get Started Today
Integrating Grounding with Parallel Web Search is simple and built directly into your Agent Platform workflow. To get started, subscribe to the Parallel service via the Google Cloud Marketplace, accept the Terms of Service, and review the pricing. Purchases count towards your eligible marketplace benefits, with consolidated billing on your Google Cloud invoice.
After subscribing, navigate to Agent Studio. From the navigation menu, click + New and select Chat. Expand the Model settings pane and click the Grounding with Partners toggle to open the Customize partner grounding menu. Select Parallel Web Search as your grounding source, apply the configuration, and your Gemini model will be instantly ready to generate responses anchored in verifiable, real-time web information.