How to build an agentic C-Suite with Google Cloud ADK and Cloud Run

Summary

Host Martin Omander introduces the episode by asking what happens when expensive executive salaries are replaced with artificial intelligence agents. Guest Anupam Phoghat, a Google Developer Expert and founder of Core-X Consulting, explains that he constructed an automated C-suite system running on Google Cloud for around ten dollars a month. The architecture enables organizations to process incoming data and automate high-level, repeated decision-making workflows across several business domains.

To demonstrate the concept, Anupam shares an application executing an Asia-Pacific market expansion strategy for a fictitious business-to-business cloud startup. Contextual data and competitor briefs are uploaded as markdown files. A central orchestrator reads the core objective and delegates domain-specific analytical tasks across multiple executive roles, including a Chief Strategy Officer, Chief Technology Officer, Chief Marketing Officer, and Chief Financial Officer. Each agent analyzes the landscape from its respective functional viewpoint, such as finding market gaps, planning positioning, or evaluating financial runway.

The system enforces a human-in-the-loop workflow before producing any conclusive action plan. As agents produce findings, the operator reviews each claim, retains the ability to alter assumptions, and either approves or rejects individual recommendations. Once the human operator completes the verification step, the orchestrator synthesizes the collective outputs into an executive brief. In the demonstration, this yields a dual-pronged strategy combining aggressive feature comparisons against one competitor with enterprise-tier support messaging against another.

Anupam points out that this architecture generalizes to many business processes beyond corporate governance, such as real estate advisory systems assessing market comparisons and interest rates or autonomous marketing teams drafting and launching campaigns. The solution is developed using Google's open-source Agent Development Kit, which handles inter-agent communication and provides evaluation tooling against golden datasets. Deployed on Google Cloud Run, the application costs nothing while idle and costs roughly seven to ten dollars per month when running five to six marketing campaigns that generate twenty ad creatives.