Part 2 – Gearing Up For The Self-Driving Business – Google Takes the Lead on Agentic AI

Agents – Part Two

Gearing Up For The Self-Driving Business

Google Takes the Lead on Agentic AI

Building upon my recent introduction to Google Agentspace, which focuses on enterprise and desktop users, I will now turn the spotlight to some of the other AI agent technologies announced in Google Cloud’s Next ‘25 event. These advancements signal changes that could reshape your business far beyond the scope of just the enterprise user.

Google is developing powerful tools that go beyond Agentspace. They help you create and deploy autonomous agents across your entire technological landscape. One such tool is the Agent Development Kit (ADK). The ADK provides a toolkit for your technical teams to develop powerful agents that interact not just with the user, but with each other. This open-source framework simplifies building sophisticated multi-agent systems while maintaining precise control over agent behavior.

ADK in Action: Automating Complex User Experiences

Think of the ADK as a set of advanced, open-source tools that empower your developers to build incredibly flexible, custom agents that go far beyond off-the-shelf products. The ADK provides the framework to create these agents efficiently, test them effectively, and deploy them reliably.

Let’s look at a real-world example of multi-agent AI in action and examine how the ADK and other innovative Google tools empower you and your organisation. In this example, multi-agent AI solutions replicate human interaction in online shopping, addressing the impossibility of having live operators for every session.

Consider a customer needing to exchange an amplifier purchased last week because it lacks sufficient volume and reverb. Upon logging in, they chat with a Dialog Manager Agent—the system’s orchestrator. The customer explains their desire to return the item and exchange it for a new one with specific features.

The Dialog Manager consults a Product Specialist Agent, which has indexed all product manuals. This agent confirms that the “Neighbour Blaster 3000” meets the customer’s specifications. The Dialog Manager then queries the Stock Agent, verifying that one unit is available.

Concurrently, a Customer Record Agent updates the customer’s profile not just with codes, but in natural language, capturing the interaction’s details, request, and sentiment. This provides richer context for human review and enables agent learning.

Finally, the Dialog Manager Agent confirms the exchange transaction and passes the details to the Order Agent, which handles subsequent status updates from the warehouse to the doorstep.

This multi-agent system efficiently manages a complex return, exchange, feature check, and stock query, updating customer records meaningfully—all while mimicking a helpful human interaction. Taking this further, it is possible to integrate features like voice from Google’s AI Customer Engagement Suite, meaning you can achieve all of this through natural, spoken conversation.

Out of the ADK box come agent-to-agent protocols which enable multiple agents to work together to achieve customer satisfaction. It simplifies building robust, autonomous agents and ensures deployment readiness with options like Google Cloud Run and Vertex AI Agent Engine. The ADK also offers flexibility and freedom, and works with many varied AI models.

What’s In It For My Business?

So much for the technology and how it works – the real question is, what do these advancements do for your business?

Increase Efficiency and Speed Across Your Business:

ADK and the agents built with it will help you automate any multi-step process across your operations and leverage parallel processing where multiple agents work simultaneously.

Improve Decision Support at Every Level:

You will be able to equip your teams with powerful, yet flexible agents that can gather information, analyze data, and provide insights directly within their workflow.

Future-Proof Your Organisation:

ADK will form a solid foundation for leveraging advanced AI capabilities across your organisation, eventually creating an “agent workforce” tailored to automate and optimize various aspects of your business operations.

Dramatically Enhance Your Customer Engagement:

Automating the workflows for handling customer inquiries, providing faster responses, offering personalized interactions (using customer history), proactive issue resolution, and even AI-driven voice communication will help you achieve better customer outcomes and higher satisfaction.

Solve Complex Operational Problems:

Tackle challenges across different departments that change frequently or require specialized knowledge by assigning tasks to dedicated agents working in concert. Free your people to meaningful and more complex tasks requiring a human touch.

Debrief: The Agentic AI Summary

The Google Agent Development Kit is a powerful, flexible toolkit, and it is available now. It empowers teams to build sophisticated, custom AI agent systems that can automate complex tasks, enhance customer interaction, and solve challenging business problems. By enabling the creation of collaborative multi-agent systems, the ADK offers a strategic pathway to harness AI for tangible business benefits and increased operational efficiency.

Agentic AI is a new way of putting technology to work when solving problems and innovating. The Agent Development Kit is the means of starting construction, but the bigger picture here is the revolution in systems thinking and how we deploy software to enhance and supplement the work we do. Agentic AI challenges us to rethink how technology can shape the organisations we are building and leading. In many ways, it is not too far-fetched to start envisioning the ‘self-driving business’.

Contact us here at Codento to discuss the potential applications of AI agents and ADK. We can show you where you could be heading with Agentic AI.

For more details on Agentic-AI or any of the other topics raised here, contact matthew.wooller@codento.com

Matthew Wooller, Data and AI Strategist, Codento