AGENTIC INTELLIGENCE

Sood and Venkatraman present a guide to improving business performance through the use of artificial intelligence.

The authors discuss a current bogeyman in public discourse: so-called “agentic” AI, programs that don’t merely respond to prompts or furnish information, but “can understand goals, plan, and take actions using tools and data to achieve them on behalf of a user.” They frame this kind of autonomous software, increasingly used in contexts from health care to business to city planning, as a critical component of the future, a tool whose use will “differentiate winners and losers.” The authors, digital business strategists with years of experience (Venkatraman as a scholar, Sood as the founder of an AI company), advocate a faster, more open process of coordinating agentic AI with business operations—even though, as they note, opening data creates concerns about risk and trust. The authors argue that trust in such systems isn’t to be assumed; it should be built into the architecture. “When trust is automated, auditable, and enforceable,” they write, “it scales.” Their key concept is “data in motion,” emphasizing the real-time updates that they believe will make AI indispensable in future applications from agriculture to self-driving cars. They engagingly bring this notion to life by opening their book with a glimpse of society in 2030, when “AI agents are now crucial coworkers (much more than tools) in most companies’ technology systems.” The authors present reams of complicated information underpinning such a vision in clear, forceful prose. Readers may feel that the text sometimes sounds like an extended brochure for Sood’s own AI company, Reltio, complete with sales-pitch lines seemingly worked over by a marketing department (“Reltio was built around a simple idea: if the business runs in real time, your data has to keep up”). Still, the consistent confidence and optimism of the authors’ positions, including the assertion that agentic AI can be an invigorating and useful part of business as opposed to a simple bogeyman, is much appreciated.

A clear and energetic guide to building “trusted data ecosystems.”