Google claims its artificial intelligence can take on the work of frontline deployment engineers

Google Cloud Vice President and General Manager Andy Gutman

Author: Kevin McLaughlin

Following Palantir's pioneering of this model, OpenAI, Anthropic, Microsoft, and Amazon have all invested billions of dollars to hire so-called Frontline Deployment Engineers (FDE) to implement new artificial intelligence technologies into the business scenarios of enterprise clients.

Google has taken a different approach: the company states that AI can automatically complete part of the work of frontline deployment engineers.

Andy Gutman, Vice President and General Manager of Google Cloud Database Products, stated that although Google Cloud recently announced the hiring of hundreds of engineers to assist clients in developing applications based on the Gemini large model, Google is now leveraging AI agents to automatically complete some of the work typically performed by professional consultants for large enterprises, primarily for enterprise data organization and sorting.

In the past, frontline deployment engineers needed to be on-site at client locations to organize proprietary internal data, enabling AI to better understand the data and help enterprise employees complete data analysis more quickly and accurately. For example, different enterprises have varying definitions of financial metrics such as "total revenue," and frontline deployment engineers relied on manual processing to ensure that AI could recognize and understand these definitional differences.

Gutman noted that Google has recognized the limitations of this heavily human-dependent model. "If you want to activate and utilize all of a company's data, you will never be able to hire enough people to accomplish this."

In his view, to maximize the value of AI agents, it is essential to open up all enterprise data to them, including various documents and legal contracts. Google's AI agents will scan and organize all client data, clarifying the relationships between data and various business processes such as sales and inventory management.

This process will generate data rich in contextual information, including knowledge graphs and semantic layers. With this foundation, the workload and costs required for AI agents to handle multi-step tasks (such as invoice processing and new employee onboarding) will be significantly reduced. This column has previously discussed these key data layers.

According to a Google Cloud spokesperson, Virgin Media O2, a UK telecommunications and media company, utilized Google AI agents to complete the integration of 20,000 independent datasets; if done manually, this work would require thousands of hours. The integrated datasets enable the enterprise's own AI agents to quickly retrieve relevant data needed for business operations.

Gutman stated that earlier this year, Palantir launched an AI version of frontline deployment engineers with some similar functionalities, but Google Cloud views this AI agent that generates business context as a competitive advantage, as it is part of Google's Knowledge Catalog data management product A Google Cloud spokesperson added that, on one hand, Google Cloud and the DeepMind team under Alphabet are collaborating to create "the most complete business context possible" for AI agents; on the other hand, unlike many competitors, Google has its self-developed Gemini large model, which can be used to troubleshoot and verify various issues that arise during the operation of the agents.

Gutmann admitted that Google's AI agents generating context are not infallible, and customers still need to configure a small number of personnel to review and verify the AI output results.

Google Cloud hopes that this process can be fully automated in the future. "At this stage, human intervention is still occasionally needed to confirm or reject the results. But as new data continues to flow in, this system can ultimately operate entirely autonomously by the agents."