Agentic AI is rapidly moving from experimentation into enterprise operations – across customer service, sales, procurement, finance and software engineering. The opportunity is compelling: faster execution, lower cost-to-serve and better decisions.
But the real challenge doesn't begin until AI agents start working across enterprise systems, business processes and teams.
At that point, success is no longer defined by what AI can do. It's defined by whether your organisation can make it secure, governed, observable and economically sustainable at scale.
That is the central point of view of Paul Fayle (Partner, Deloitte Digital) in "The Foundation of the Agentic Enterprise" — a report that asks what capabilities it really takes to move agentic AI from pilot to enterprise scale, using MuleSoft from Salesforce as a worked example.
As AI agents become more autonomous, they introduce a new class of enterprise challenge: opaque decision-making, machine-speed failures, uncontrolled costs and fragmented governance.
Scaling agentic AI isn't simply about deploying more agents. It requires the right integration platforms, architectural patterns and operating model to orchestrate, secure, monitor and govern agents across SaaS applications, data platforms and legacy systems.
Because the organisations that lead the agentic era won't just build smarter agents.
They'll build stronger foundations.
Paul Fayle is a Partner and national Salesforce CTO within the Australian Salesforce Practice of Deloitte Digital. He has over two decades of experience in enterprise and solutions architecture, and delivery of complex enterprise projects spanning a diversity of industries, with an emphasis on customer relationship management (CRM), master data management (MDM), and data quality. Paul is also a Salesforce Certified Technical Architect, the highest qualification available on the platforms, and is one of only a few hundred globally to hold this certification.
