AI moving into enterprise workflows

Artificial intelligence is moving from pilot projects into corporate operating systems, with Google DeepMind now enlisting top consultancies to roll it out inside large companies. (deepmind.google) On April 22, Google DeepMind said it had partnered with Accenture, Bain & Company, Boston Consulting Group, Deloitte, and McKinsey to help clients adopt “frontier” AI and redesign work around software agents. DeepMind said only 25% of organizations have moved AI into production at scale so far. (deepmind.google) A software agent is an artificial intelligence system that can plan steps and carry out tasks, not just answer a prompt once. DeepMind said its new Deep Research and Deep Research Max agents, released April 21, can work across the web or custom sources, support Model Context Protocol connections, and generate visualizations as part of long research jobs. (blog.google) Google said the upgraded research agents are built on Gemini 3.1 Pro and follow a December 2025 release that first exposed Gemini Deep Research to developers through the Interactions API. The company is pitching them for “long-horizon research workflows,” meaning projects that require many searches, source checks, and synthesis steps over time. (blog.google (blog.google)) The consultancy deals arrived alongside Google Cloud Next ’26, where Google framed the “agentic enterprise” as a current deployment phase rather than a lab concept. Google said nearly 75% of Google Cloud customers now use its artificial intelligence products, and 330 customers processed more than 1 trillion tokens each over the past 12 months. (cloud.google.com) (blog.google) Google also used the event to launch Gemini Enterprise Agent Platform, which combines Vertex AI model tools with agent integration, security, and operations features. The company described it as a centralized system for building and managing autonomous agents across an organization. (blog.google) The sales pitch is shifting from standalone chatbots to systems that can search internal data, call software tools, and complete multi-step work inside existing business processes. Google Cloud said in a March 26 post that companies are moving from chat interfaces toward agents that automate complex workflows. (cloud.google.com) Google is not the only company pushing that model, but its latest announcements tie together three parts of the market at once: model research from DeepMind, cloud infrastructure from Google Cloud, and implementation services from firms that already run large corporate transformation projects. That combination gives chief information officers a familiar procurement path for adopting newer AI systems. (deepmind.google) (cloud.google.com) The immediate test is whether these agents can move beyond demos and into routine work without creating new security, accuracy, and oversight problems. Google’s answer, at least this week, is to package the models, the tooling, and the consultants together and sell AI as workflow infrastructure. (blog.google) (deepmind.google)