Across global enterprises, AI is no longer a future vision; it is already in employees’ daily workflows. Yet despite this widespread use, most organizational AI initiatives still struggle to cross the chasm from pilot to production.

In a panel hosted by Calcalist’s Maayan Manela, Boaz Ziniman, Principal Developer Advocate EMEA at Amazon Web Services (AWS), shared a striking data point. Around 95 percent of employees across organizations already use AI in some form, but between 50 and 80 percent of AI projects never make it into production environments. The problem is not a lack of experimentation; it is a lack of operationalization.

Ziniman emphasized that for AWS and Amazon, AI did not begin with GenAI. It has been embedded in products, logistics, and customer experiences for more than two decades. The difference today is scale. To move beyond pilots, organizations need strong data foundations, clear governance, and executive sponsorship that ties AI directly to business outcomes.

On the second day of AI Week by Commit and Calcalist, in collaboration with AWS, Ofer Feldman, Co-Founder and CTO at Stampli, and Ronni Zehavi, Co-Founder and CEO at HiBob, brought this perspective to life. Stampli has built an AI-first architecture across its procure-to-pay workflows, with its bot Billy acting as a recognizable teammate rather than a distant back-end feature. HiBob, which serves more than a million users, adopted a dual approach that combines top-down leadership with bottom up experimentation, resulting in 2,500 internal agents and 95 percent organizational adoption.

Throughout the discussion, one message was constant. The companies succeeding with AI are those that treat data as a strategic asset, invest in readiness, and design AI with trust, user experience, and measurable productivity at the center. Enterprise AI is no longer about proofs of concept; it is about building production-grade systems that can scale reliably and responsibly.

‍