What does it actually take to move AI from experimentation to production at enterprise scale? At Google Cloud AI Live in Amsterdam, we sat down with Asaf Salam, Head of AI Benelux, Italy & Israel at Google Cloud. He has a good perspective on AI adoption in different markets and where organizations are struggling. What does a structured, strategic approach to AI adoption really looks like in practice?
Salam explains how a common misconception divides leadership teams from individual contributors: executives want AI to help consult on strategy, while employees just want to automate daily tasks. Both are valid, but without a unified platform and a clear operating model, organizations end up with a patchwork of disconnected tools that become unmanageable. Google Cloud addresses this with a new end-to-end AI organization, including Forward Deployment Engineers (FDEs) who work on-site with customers to scope, build, and ship AI products fast.
The conversation goes deep on tokenomics, the planning discipline around token usage that too many organizations overlook, and gives practical guidance on when to use Gemini Flash versus Gemini Pro. Salam also walks through the Agent Designer (v2), which lets non-technical employees build, test, and distribute their own agents across an organization with no code. Whether you’re just getting serious about AI or trying to rationalize a fragmented AI landscape, this interview is packed with actionable insight.
• The perception gap between leadership and individual contributors in AI adoption
• Why consumer AI experience does not translate directly to the enterprise
• How Google Cloud’s Forward Deployment Engineers (FDEs) accelerate production deployments
• What tokenomics means and how a 3-tier provisioned throughput model works
• When to use Gemini Flash vs. Gemini Pro for different task types
• How Agent Designer v2 enables no-code agent building at enterprise scale
• Why planning, not building, is the most critical first step
• Salam’s view on whether AI adoption will slow down or keep accelerating
Chapters:
0:57 Leadership vs. employees: the AI perception gap
2:19 From AI pilots to strategic platform: where to start
4:59 Google’s new AI org and forward deployment engineers
6:25 How Google Cloud and customers divide responsibilities
7:20 Consumer AI vs. enterprise AI: a multidimensional shift
8:28 Planning for AI: capacity, tokenomics, and token budgets
11:09 Choosing the right model: Gemini Flash vs. Pro
13:35 Gemini Enterprise and no-code agent building
15:06 Will AI slow down or keep accelerating?
Google Cloud AI, Gemini Enterprise, enterprise AI adoption, AI agents, tokenomics, Forward Deployment Engineers, Gemini Flash, Gemini Pro, no-code AI, Agent Designer, AI strategy, Google Cloud Platform, AI governance, AI security, multimodal AI