Everyone’s building AI agents. Far fewer are scaling them. 

In this episode of Beyond the bot, recorded at Capgemini’s Barcelona office, we go deep into the gap between a promising proof-of-concept and a production-ready agentic system – and what it actually takes to close it. 

Joining us are guest speaker Rowan Curran, Principal Analyst at Forrester Research specialising in AI applications and data science; Raul Bartolome, AI Applied to Industry Global Portfolio Lead, Capgemini; and your host Itziar Goicoechea, Senior Director, Agentic AI Offer Lead, Capgemini.  

Together, they bring a rare combination of industry-wide research, client delivery experience, and hands-on technical leadership – and between them, they’ve seen what works, what stalls, and what enterprises consistently underestimate when scaling agents beyond the pilot stage. 

Together, we cover: 

  • Why getting started with agents is easy – and scaling them is not 
  • The data, infrastructure, and governance foundations you need before writing a single agent 
  • How to think about agent identity, security, and dynamic access policies 
  • Why evaluation and testing is the industry’s most underinvested problem 
  • Real use cases gaining traction: document intelligence, invoice processing, IT support automation, loan pre-approval, risk analysis 
  • The human-in-the-loop question – when to let agents run, and when humans must stay in control 

If your organisation is asking “how do we move from pilot to scale?” – this conversation is for you.