- Google, Microsoft and OpenAI are all driving a shift from AI assistants to proactive AI agents
- These agents promise to bring more intuitive interactions with complex data systems
- But fully autonomous AI systems are still a ways off
Software companies are exploring a new paradigm: agentic artificial intelligence (AI), and they’re using it for better data analytics and business intelligence. The broader shift to AI agents “signals a change in AI perception and maturity,” said Alexander Wurm, principal analyst at Nucleus Research.
Unlike traditional AI assistants designed to follow commands, agentic AI systems are built to proactively analyze, recommend and even take actions on behalf of users. While organizations “still value AI assistants delivered by technology vendors,” Wurm said they’re increasingly experimenting with custom use cases using API-enabled AI services.
Google Cloud, for example, is privately trialing its Looker, a conversational data analytics agent that connects with popular workplace tools like Slack, Microsoft Teams and Google Chat so users can interact with data within the platforms they already use for communication.
Tableau, a Salesforce subsidiary, was an early adopter of proactive AI capabilities, rebranding its AI assistant as an agent. Meanwhile, OpenAI announced plans to launch its own AI agent as a research preview and developer tool in January, per Bloomberg. And smaller companies like ThoughtSpot are also entering the space with tools like Spotter, their agentic AI analyst.
According to Wurm, first-mover advantage won’t define the market — “rather, solutions that best connect underlying data with semantic contexts and business logic will see the greatest adoption." In that context, he said vendors like Google, Microsoft and Salesforce are "particularly well-positioned” for success.
“Several other vendors have agentic products, but many are still in public or private preview,” Wurm told Fierce Network. Nucleus Research expects to see a proliferation of agentic solutions next year from “the vast majority” of software providers.
The driving forces behind agentic AI
At the heart of this transition is the need for more intuitive interactions with complex data systems.
“GenAI has captured the imagination of all the amazing applications that could be built,” Peter Bailis, VP of Engineering at Google Cloud, told Fierce.
Google Cloud's Looker agent lets users query and analyze data using natural language. Instead of needing to know complex query languages or navigating dashboards, users can ask questions like, "What were last month's sales figures?" and get immediate insights.
For users, the rise of AI agents shifts the focus from answering basic questions to driving strategy and outcomes. “It’s exciting,” Bailis said. “Imagine asking complex, multi-step questions of private data and getting meaningful answers. That’s the long-term unlock in the generative AI and data space.”
Bailis said the Looker AI agent has “a handful of really promising use cases” in the telco space. He highlighted three key use cases: instant detection and remediation, customer service and business performance.
The road ahead
Fully autonomous AI systems are still a ways off. Tools like Looker’s conversational analytics still keep humans in the loop, Wurm noted, delivering insights while leveraging Google’s data infrastructure and semantic knowledge.
Indeed, Bailis explained that while agents like Looker’s are evolving beyond simple Q&A models like the first renditions of ChatGPT or Gemini, the industry is still “one or two model generations” away from seamless agent-to-agent collaboration. “Even with current models, there’s so much more to build,” he said. “These conversational analytics APIs and agents are just scratching the surface."
However, Bailis said that the progress made thus far with AI agents will already have been powerful. “If we stopped today, their impact in the data space would still be massive—on par with cloud computing,” he said.
Bailis summed up the potential of agentic AI in data analytics: “Every business cares about three questions: What’s happened? Why is it happening? What’s happening next? It’s very painful to get those answers today. With this conversational data agent, once you define those metrics, you’ll be able to get those answers over and over again.”