Artificial intelligence tools have become a key part of businesses. While tech giants like Microsoft, Anthropic and OpenAI continue to spend on forward-deployed engineers (FDEs), executives who work with businesses to help them use AI effectively, Google may take a different approach. According to a report from The Information, Google is developing AI agents that can automate part of the work these specialists usually take care of.
Google Turns To AI Agents
According to the report, Google Cloud vice president Andi Gutmans said the traditional approach has a major drawback. He reportedly said, “If you want to move to activating 100 per cent of your enterprise data, you’re not going to be able to hire enough people to make that happen.”
Rather than fully relying on human engineers, Google is allegedly using AI agents to examine large chunks of company data. Reportedly, these systems can analyse files, contracts and other business information to understand how various pieces of data connect.
According to Gutmans, the goal is to create richer context for AI systems. This may include knowledge graphs and semantic layers, which could help AI agents to execute difficult tasks.
Notably, Google Cloud recently said it planned to hire hundreds of FDEs to help customers develop applications using its Gemini AI tools. However, the company also reportedly believes AI agents can reduce the amount of manual work required. Reportedly, Virgin Media O2 has already used Google’s agents to connect around 20,000 separate sets of data. According to a Google Cloud spokesperson, doing the same work manually would have taken thousands of hours. The connected information allows the telecommunications company’s own AI agents to locate relevant data more easily.
Google admits the technology is not yet perfect. Human employees still need to check the AI’s work and correct mistakes when necessary. However, the company expects that requirement to shrink over time. Gutmans said, “There’s a period of where humans will occasionally step in to say yes or no. But over time, it can become fully agentic as new data comes in.”
