AI agents reflect a broader industry trend: the shift from bots and copilots to autonomous, task-oriented agents that can not only predict and generate responses but also take action. Based on our interaction with over 60 industry leaders, including C-level decision-makers, advisors, providers, and tech vendor executives, 88% of enterprises are expected to integrate synchronous AI agents into their operations in some capacity in 2025. However, only 10% will scale the deployment of these agents across more than 40% of their workflows, focusing on process automation and predictive insights.
At Google Cloud Next 2025, the narrative of The Wizard of Oz took on a new dimension— not just as a nostalgic reference but as a literal reimagination with generative AI. Sundar Pichai, CEO of Google and Alphabet, announced their collaboration with Sphere Studios on a generative AI-powered reinterpretation of the classic film The Wizard of Oz, set to premiere on August 28 at the Las Vegas Sphere. Its premise—reimagining a traditional journey through a modern, immersive lens—mirrors the direction Google is taking with its cloud strategy.
Just as Dorothy’s black-and-white world gave way to a vibrant technicolor Oz, enterprises are moving from conventional cloud environments to modern cloud platforms where AI is not an add-on but a foundational layer. This year’s announcements, from Gemini-powered multimodal models to enterprise agents, signal Google’s continued intent to guide organizations along this transformative path—a new kind of yellow brick road built on open ecosystems, secure infrastructure, and AI-first innovation.
Redefining Creativity with Generative Multimodal Models
Google announced advancements across the following suite of models, including text-to-image, text-to-video, image-to-video, and text-to-music, enabling enterprises to expedite content creation activities:
-
- Imagen 3, the text-to-image model in Google’s portfolio, can now reconstruct missing or damaged portions of an image and provides object removal capabilities.
- Veo 2, a text-to-video model, can produce several minutes of 4K content, equipped with editing tools such as dynamic inpainting, outpainting, interpolation, and camera shot composition. This offers creative control to users without requiring sophisticated prompting or specialized expertise.
- Lyria, the text-to-music model, can transform text prompts into 30-second music clips and create compositions across various musical genres. These are helpful for enterprises to create soundtracks for marketing campaigns and product launches to improve brand experiences.
- Chirp 3, which synthesizes human-like voice from 10-second input, helps enterprises weave AI-powered narration into existing recordings.
Kraft Heinz, Agoda, and L’Oreal Group are some companies that have generated diverse cinematic shots and videos for their marketing campaigns using the abovementioned tools.
However, the rapid adoption of generative AI has elevated responsible AI as a strategic imperative, as it introduces new risks, such as data security, IP misappropriation, and hallucinations, while amplifying existing concerns around ethical bias and data privacy. This has led to platform vendors enhancing their responsible AI solutions by integrating advanced generative AI guardrails to address these evolving challenges.
Based on Avasant–nasscom Digital Enterprise 5.0 survey of 550 enterprise executives, 46% of enterprises are dedicatedly investing in responsible AI, out of which 34% are cross-validating AI output with responsible AI guidelines internally, as shown in Figure 1 below.
Figure 1: Responsible AI initiatives implemented by enterprises
Keeping with this trend and enterprise demand, Thomas Kurian, CEO of Google Cloud, emphasized Google’s usage of responsible AI practices with built-in precautions across the multimodal models, such as digital watermarking with SynthID technology, safety filters, data governance, and indemnification support where Google will protect its customers for third-party IP claims and copyright issues against content created with generative AI.
AI Agents Take the Center Stage in the Next Wave of Innovation
A central theme throughout the event was the rise of AI agents as the next paradigm in enterprise software. The adoption of AI agents is no longer limited to a single task; rather, agents are dynamic, multifunctional digital coworkers capable of reasoning, dialog, and execution. At the event, Google introduced an array of enablers to foster the development of these agents. Google Agentspace provides employees with advanced Google models, quality search, AI agents, and enterprise knowledge. The Agent Development Kit (ADK) simplifies multiagent creation and allows developers to build complex agent behaviors with minimal code. Supporting this is the Agent2Agent (A2A) protocol, a new standard for inter-agent communication regardless of the underlying technologies. The Agent Garden is a repository of prebuilt tools, connectors, and samples, helping developers seamlessly plug into existing systems such as Apigee APIs and cloud databases. More than 50 major partners, including Accenture, Salesforce, Deloitte, SAP, ServiceNow, and TCS, are collaborating to define this multiagent ecosystem.
Based on our research on synchronous AI agents, 90% of synchronous AI agent use cases focus on the backend and diagnostic processes. IT, sales, and marketing functions lead in synchronous AI agent experimentation due to their maturity in deploying generative AI agents and proven productivity gains. Figure 2 illustrates key use cases across different business functions and the ones highlighted in bold represent use cases that are most ripe for synchronous AI agent architecture in the next six months.
Figure 2: Synchronous AI agent use cases being adopted by enterprises
Aligning with this market trend, Google introduced various agents that are creating high business impact for clients. In addition to exciting announcements around Google Agentspace, which helps scale the adoption of enterprise search and AI agents across the enterprise, Google rolled out Customer Agents for customer service capabilities, Creative Agents for marketing campaigns, Data Agents for unlocking and accelerating data insights, Security Agents for faster detection and response, and Coding Agents for software development.
The Way Forward
The leading cloud providers are evolving beyond traditional infrastructure support to cater to customers’ growing demands, with generative multimodal models, AI agents, and responsible AI practices becoming standard offerings. Effective AI agent deployment requires strong governance to manage process complexity and business alignment.
Organizations must establish clear guidelines for data access, model version control, and AI agent performance tracking to prevent inconsistencies in outputs and deviations from business objectives. A structured change management process should cover model updates, workflow adjustments, and operational changes to ensure smooth transitions and minimal disruption.
By Dhanusha Ramakrishnan, Lead Analyst, and Gaurav Dewan, Research Director, Avasant