Now on Packt · Publishing September 2026
Building the Agentic Enterprise on Google Cloud
A Practical Field Guide to Designing, Deploying, and Operating Agentic AI Systems.
AI Agent adoption usually starts in one business unit, a Proof of Value or Concept as they would call it. Assuming AI delivers on its magic and wonders - as it typically does and this use case proves itself, before you know it the next immediate ask is to scale, and get this into Production. The challenge is that to scale agentic architecture means the wider enterprise needs to have a strategy to scale beyond this single business unit. That very transition is the dichotomy you have to bridge and get it right or this poc value never materialises in effect.
You see, as a platform engineer for a long time, I’ve had a developer itch on the side. Getting hands-on with AI and AI use cases as they came onto the scene to prove value-add from hype first-hand. I have worked in lean startups and Linux shops building and orchestrating large enterprise platforms into production. And this advent of technology still tends to fall back on Human Alignment challenges of People and Processes first and foremost. I’ve been proving it out, building and architecting hands-on, but also figuring it out and unpacking these challenges with my industry peers on how to effectively transition from AI Native, even Agentic Builder-Developer, to AI-transformed Organisation Agent Platforms as a Technology and Business Operator.
I hope this book helps you find where agentic AI makes the most sense for your organisation, and where the real value is.

Jaroslav Pantsjoha
Technical Director · Google Developer Expert, AI/ML

What you’ll take away
- →Why most agentic pilots stall before production, and how to get real value from your agentic AI investment
- →When to split work across many agents and when to keep it to one, with the right model and design for the use case
- →Getting AgentOps and EvalOps right as disciplines, the key to production-grade agentic systems
- →Designing security and identity into agents from the start
- →How the operating model changes when one expert runs a squad of agents
Who it’s for
The person who already has the pilot, the pressure, and the executive attention, but not yet the operating model to carry any of it.
- · Enterprise Architects
- · Technical Directors
- · Platform Leads
- · CTOs
- · Engineering leaders accountable for production
What’s inside — 12 chapters, four parts
Part 1 — Foundations
- Why Agentic Systems Fail to Go Live
- The Agentic AI Foundation
- Single-Agent and Multi-Agent Systems
Part 2 — Designing
- Core Architectural Patterns for Agentic Systems
- Multi-Agent Orchestration
- Context, Memory, and Knowledge
Part 3 — Building & Operating
- Security, Identity, and Guardrails
- Building and Deploying Agentic Workflows
- State, Memory, and Decision Persistence
- AgentOps and EvalOps
Part 4 — Scaling
- Platform Engineering for Agentic Systems
- Operating Models and Business Transformation
About the author
Jaroslav Pantsjoha is an enterprise architect and technical advisor. He designs cloud and agentic AI platforms for large organisations and advises the executives backing them.
Full bio →Now on Packt Publishing, also on Amazon.