Google Cloud debuts AI Agent Space to take on Microsoft and Salesforce

The enterprise cloud wars have officially moved past the stage of bragging about model parameters. Now, it is all about who can build the most bustling app store for AI agents.

Google Cloud just unveiled AI Agent Space, an ecosystem initiative paired with a dedicated storefront on the Google Cloud Marketplace. The goal is straightforward: convince third-party developers to build task-automating agents on Google Cloud, and give corporate buyers a single hub to deploy them.

A marketplace with big promises and empty shelves

At launch, AI Agent Space feels more like a soft opening. The catalog currently features just over a dozen prebuilt agents, including marketing assistant Typeface and ThoughtSpot's Spotter, a natural language business analytics agent.

Kevin Ichhpurani, president of Google Cloud's global partner organization, stated that hundreds of additional agents are slated to join in the coming months. To sweeten the deal for developers, Google is dangling perks like direct engineering support, early access to unreleased AI features, and co-selling opportunities backed by Google's marketing machine.

Oddly enough, Google spent much of its launch announcement flexing impressive internal solutions built by consulting giants like Deloitte, Bain, Accenture, and PwC. Deloitte's Care Finder agent, for instance, cuts patient provider search times from eight minutes down to under sixty seconds, while Bain claims a 15% efficiency boost for wealth managers.

The catch? Almost none of these high-profile enterprise implementations are actually available to purchase on the AI Agent Space yet. Google essentially used them as proof of concept, hoping to entice other businesses into commercializing their custom internal tools as SaaS products on its platform.

The enterprise squeeze from Microsoft and Salesforce

Google is not launching this in a vacuum; the timing smells like a defensive play against mounting pressure from rivals.

Microsoft just dominated headlines at its Ignite conference, revealing that over 100,000 organizations already use Copilot Studio to build AI agents. Microsoft also flexed an arsenal of more than 1,800 models in Azure and an "agent mesh" architecture designed for multi-agent workflows. Meanwhile, Salesforce is aggressively rolling out Agentforce, and SAP is embedding collaborative Joule agents deep into enterprise ERP and supply chain pipelines.

Google has offered developer tools like Vertex AI Agent Builder since April, but until now, it lacked the commercial marketplace layer needed to turn bespoke experiments into scalable enterprise purchases.

Why it matters

For enterprise tech teams, AI Agent Space is worth watching, but probably not worth rushing into just yet. If you are an enterprise developer building domain-specific automation, joining Google's program could be a lucrative route to tap into enterprise cloud budgets through co-selling. But if you are a buyer looking for plug-and-play business agents today, the catalog is still too thin to replace bespoke development.

On an industry level, the center of gravity has shifted from raw model benchmarks to packaging and distribution. Cloud giants are realizing that most companies do not want raw APIs; they want software agents that solve specific operational headaches.

We think Google's partner-first strategy is sensible, but opening a digital storefront with barely a dozen offerings while parading unavailable consulting projects feels a bit rushed. Microsoft already has a massive head start with Copilot Studio's enterprise adoption. Unless Google can rapidly convert its ecosystem promises into tangible, ready-to-buy software, it risks losing the high-margin enterprise agent race to Redmond.