
Companies trying to stay ahead in analytics, efficiency, customer experience, and overall competitiveness are leaning harder on AI to run their operations. Across industries including manufacturing, logistics, retail, and financial services, I’m seeing enterprise data platforms become the place where day-to-day operational inputs (inventory updates, customer activity, equipment readings, order changes, supplier signals) actually land, and where teams make real decisions from that data. These platforms are also starting to take on routine actions like flagging delays, adjusting forecasts, routing work, or prompting the right person when something needs attention. What stands out to me in Google Cloud’s latest updates is how they’re pushing AI deeper into the data platform so enterprises can use their information more effectively — organizing it better, interpreting it faster, and turning it into steps that help run the business.
Google isn’t alone. Microsoft is pushing hard on semantic models and agent frameworks through Fabric and MCP. AWS is doubling down on integrating AI directly into its data, analytics, and operational pipelines. IBM is weaving AI into integration, governance, and hybrid data estates through watsonx.data, StreamSets, and webMethods. Oracle continues to ground its strategy in tightly coupled data, analytics, and application logic, with AI agents emerging inside Fusion. ServiceNow is embedding AI into workflows and knowledge models that align more closely with how employees actually work. Each company is moving in a similar direction, with AI becoming the new operating layer of the enterprise, but their starting points, architectures, and strengths differ in meaningful ways.
At the center of Google’s shift is the company’s Data Cloud — framed as an “intelligent operating system” for the enterprise — that brings automation, multimodal AI, and governed intelligence together on a global infrastructure capable of reasoning across data in real time. This evolution moves Google Cloud from data-driven to agent-driven, with many potential impacts for how companies act on data.
AI as a Business Imperative
For all the hype around AI over the past three years, manual data work and disconnected systems still slow down decisions in most companies. Google’s approach to solving this challenge is to use AI agents that automate routine steps, work across disparate data sources, and connect with tools that companies already use. Far from acting like the one-off chatbots from the earlier days of generative AI, these agents are set up to understand business context and plug into existing systems and workflows.
This vision is now translating into production impact. Companies are starting to use Gemini across Google Workspace and Google Cloud to handle routine tasks like summarizing data, generating reports, and building plans. By offloading repetitive work to agents, people can stay more focused on real business problems instead of chasing down details in spreadsheets and pipelines. As one example of this offloading, Priceline uses Gemini to improve trip planning and deliver more personalized travel recommendations for both consumer and enterprise users. The goal is to reduce the time spent on manual data steps so users can move from analysis to decisions faster.
Google’s New Releases — Building Blocks for Agentic Workflows
Google Cloud’s October 2025 releases expand this foundation through Gemini Enterprise, a unified agentic platform, and a full suite of data-focused agents that bridge operational and analytical domains. Gemini Enterprise is designed as a controlled environment where analysts, engineers, and business users can build and run AI agents, create automations, and connect to data sources in one place — without needing deep technical skills. It also ties directly into tools people already use; this includes Google offerings such as Google Workspace, BigQuery, and Vertex AI, but it also extends to third-party products like Microsoft 365 and Salesforce. In each case, the point is to enable users to act on data without jumping between systems.
Another key update is the deepening of AlloyDB AI, which now includes advanced vector search and integration with AI models, allowing developers to build generative AI applications directly on their operational transactional data.
Google is also pushing agents deeper into the data stack, not just as helpers, but as working parts of the analytics pipeline. The Data Engineering Agent in BigQuery handles ingestion, cleanup, and transformation through natural-language instructions, removing much of the repetitive scripting that slows data teams. The Data Science Agent links BigQuery with Vertex AI to manage complete machine learning workflows, from exploration to deployment, without constant tool switching. For analysts, the Conversational Analytics Agent now includes a Code Interpreter mode, making it possible to run real Python analysis directly inside Looker or BigQuery and get visual, explainable results without building notebooks.
Google also gives developers and system integrators the tools to build their own agents, not just use prebuilt ones. The Gemini APIs and Agent Development Kit (ADK) let teams create and deploy custom agents that fit specific industry, security, or compliance requirements. These frameworks include built-in governance support through Model Context Protocol (MCP) servers and Dataplex metadata, so agents inherit company rules from day one. Agentspace, now a capability within Gemini Enterprise, enables organizations to manage AI agents at scale. Customers can use a centralized platform to find, publish, monitor, and govern AI agents across their organization, similar to an internal app store.
Gemini Enterprise also incorporates audit trails and policy controls to ensure transparency and compliance. Teams can register agents, track usage, and enforce role-based access to sensitive workflows. This gives IT and security teams a unified view of agent behavior and dependencies, helping to prevent the spread of shadow AI projects. Agents are also reviewed for security and interoperability to help ensure reliability and compatibility.
Why It Matters — Changing the Economics of Data
Taken together, if enterprises truly embrace Google’s approach with these new capabilities, it could change the economics of working with data. For years, the lion’s share of analytics budgets has been spent on unglamorous work — data prep, pipeline cleanup, quality checks — often consuming 70% to 80% of total effort. Agent-based workflows cut down those costs by automating ingestion, integration, and reporting — and that’s before we get to AI-specific tasks such as model training. BigQuery’s new continuous intelligence features also help shrink the time from query to decision, often dramatically, by harnessing real-time data instead of using static data. Every agent also inherits Google Cloud’s security and governance settings so that authentication, lineage, and privacy policies can stay consistent.
Google touts the measurable benefits for CIOs and CFOs, including lower consulting spend, reduced data backlog, and faster analysis without increasing cloud costs. The governance layer matters for these decision-makers, too, because automation doesn’t work in regulated industries unless it’s auditable and traceable.
That said, what Google is offering isn’t all plug-and-play. Companies must still validate agent accuracy, set clear rules for what agents should and shouldn’t do, and take steps to control costs as the use of agents increases. Governance frameworks built into Gemini Enterprise should certainly help with this, but execution discipline will ultimately decide whether agentic AI scales effectively in the enterprise.
Industry Outcomes in Practice
Enterprise use of Gemini and Google Cloud’s AI tools is starting to show real traction across industries. In financial services, banks can use AI agents to automate parts of reconciliation and regulatory reporting to reduce both manual review time and common reporting errors. In healthcare, Quest Diagnostics uses Google Cloud AI to speed up medical research and operational analytics by connecting diagnostics and logistics data in real time. In manufacturing, Mercedes-Benz uses Google Cloud AI to optimize production quality and factory operations with data-driven automation. In retail, Target uses AlloyDB AI and semantic search to improve product discovery and personalize shopping experiences.
Supply chain teams are also starting to connect sales, inventory, and logistics signals, for instance using Gemini with Vertex AI and BigQuery to improve forecasting accuracy and reduce waste. Sports organizations have also become more involved; for example, Major League Baseball uses Google Cloud AI to analyze fan sentiment and improve real-time event operations. (For more on this, see my discussion with Rob Engel, senior vice president of baseball and content engineering at MLB.) Among other benefits, this range of applications illustrates the emerging truth that agentic AI is shifting enterprise intelligence from descriptive dashboards to interactive, outcome-driven collaboration.
Competitive Landscape — Positioning in a Crowded Market
All of these moves from Google come in the context of increased competition among AI data platforms. Microsoft is pushing Fabric and Copilot, AWS has Bedrock and SageMaker, IBM is using watsonx across data, governance, and model lifecycle automation, Snowflake is moving fast with Cortex and Polaris, and Databricks is betting on Mosaic AI. All of them are working to connect analytics with AI, but each is coming at it differently.
Google Cloud’s angle is to build around scale, openness, and interoperability. Its support for Apache Iceberg, BigLake, and open model interfaces gives customers flexibility instead of locking data into one place. The ability for agents to work with other agents across systems via the Google-developed Agent2Agent protocol (A2A) is a differentiator, too, because it lets automation extend across different business apps rather than staying trapped inside one product.
Google is also extending this strategy into industrial AI. Its partnership with NVIDIA brings Blackwell GPUs, digital twins through Omniverse, and robotics simulation with Isaac Sim into the picture. This means, for example, that manufacturers and other industrial companies can simulate physical operations before deploying AI agents on the shop floor. It’s a practical move that connects Gemini Enterprise’s data intelligence layer with real-world execution, rather than limiting AI to software workflows.
Execution Will Define Leadership
Google Cloud’s momentum is real, as shown by the numbers. It reported $15.2 billion in Q3 2025 revenue, up 34% year over year. Google Cloud’s backlog grew to $155 billion, driven by rising demand for AI-ready infrastructure and generative AI services. More than 70% of its cloud customers now use AI products, indicating that AI adoption is becoming more integral to their businesses. Google also saw a significant increase in AI-driven data workloads over the past year, with continued momentum from AlloyDB AI, reflecting a shift towards databases and data platforms designed for AI support.
Google has done a good job solidifying its data foundation and making AI easier to use. There’s proof the tech works, but the real question is whether early wins turn into long-term ROI for large, complex enterprises. What I’m watching now is whether customers can sustain the gains with faster cycles, less manual work, and clearer decisions as agent-driven workflows move into the core of operations. If customers can show measurable impact from Google agents in production, the company will have a credible case that its agentic data platform delivers real business value. Ultimately, Google’s ability to lead this shift will come down to execution.
