Secret Agentspace: Google announces new AI tool to help enterprises turn silos into lakes

Google Search is an incredibly valuable resource, to the point that it has effectively defined the modern internet. But as we’ve all experienced, Search doesn’t always get the right results. We’re often left looking for answers, despite the search engine providing many choices to dig deeper into.

Sometimes, that outcome is because the search engine has been gamed, and less-than-relevant results are presented because of SEO activities from website operators.

Also: Google’s Gemini 2.0 AI promises to be faster and smarter via agentic advances

Sometimes less-than-satisfying answers are presented because the information is unavailable online, or search strings are imprecise or ineffective. And sometimes less-than-satisfying answers are presented because the data is locked behind proprietary firewalls.

Google seeks to solve this final problem with Google Agentspace, its agentic AI offering for the enterprise. In a blog post today, Google Cloud AI VP and General Manager Saurabh Tiwary describes AgentSpace by saying, “It unlocks enterprise expertise for employees with agents that bring together Gemini’s advanced reasoning, Google-quality search, and enterprise data, regardless of where it’s hosted.”

Let’s deconstruct the core components of this offering: AI, search, and data.

Let’s start with data. It’s not just data, says Tiwary. It’s data “regardless of where it’s hosted.” That approach means it’s possible to use Google’s powerful search and retrieval tools to scan a wide variety of silos. Google mentions Confluence, Google Drive, Jira, Microsoft Sharepoint, and ServiceNow, but implies there will be more.

When combined with Google Search, this functionality effectively turns data silos into a data lake, making that information accessible across the enterprise and in cross-functional uses without regard to where it was originally captured and stored.

Also: 5 reasons why Google’s Trillium could transform AI and cloud computing – and 2 obstacles

Obviously, there are security considerations. Agentspace uses Google’s Secure by Design architecture, which provides granular control over data and who can use it. Google lists IT controls that will be applied to Agentspace, including:

• Role-based access control – Allows operators to tier permissions based on user role
• Virtual private cloud service controls – Provides data access controls within virtual private clouds
• Identity and access management integration – Provides identity management and access with granular permissions, often at the record or field level
• Customer-managed encryption keys – Customers, rather than Google, can manage their encryption keys

So, you’ve got data hosted across several enterprise and cloud infrastructures that Google’s search prowess can index and retrieve. Now, add agentic AI to that mix.

Also: Gemini’s new Deep Research feature searches the web for you – like an assistant

Earlier this week in a blog post, Google and Alphabet CEO Sundar Pichai described agentic AI as a user interface with “action-capabilities”. He listed three criteria that characterize agentic AI which can be boiled down to situational awareness, the ability to plan steps, and the ability to take action.

Awareness, planning, and action. Combine those with the ability to search across enterprise silos and you have a potent recipe for potential solution development.

Programmers have been building apps for years now that have crossed silos, operated according to algorithmic steps, had situational awareness, and taken some actions. But we’ve had to build those things using microservices and data interchange protocols, and all the tedious and complex tools of interoperability.

What if some of those internal enterprise projects no longer took a year to field, but a few afternoons of prompting and refining? That’s some powerful stuff right there.

Also: AI is moving undercover at work in 2025, according to Deloitte’s Tech Trends report

There will be work for coders to complete, especially when building and fine-tuning these agents. But the time it takes to do all the tedious interconnection stuff may be handled behind the scenes by the data robot. 

I talked about how to think about programming with AI in my 25 tips article. A big part of the message was to leave the common knowledge work to the AI and do the unique, proprietary stuff in code.

With agents, the same is true. You’ll have to describe the business processes and the things you need to see from the data. But you won’t have to bang your head against the wall to write code to extract mixed-mode data from Sharepoint, for example. Skipping that stage will save weeks.

There’s another interesting part to this announcement: integrating your enterprise data with NotebookLM. I discussed NotebookLM earlier this year when I showed how the tool could turn an article into a compellingly realistic podcast.

Also: OpenAI rolls out Canvas to all ChatGPT users – and it’s a powerful productivity tool

But NotebookLM is more than a podcasting novelty. Think of it as a tool for processing collections of documents. Users can upload batches of documents to individual folders called Notebooks. Then, they can use Google Gemini to perform AI-enabled tasks on those notebooks, whether summarizing the information, synthesizing it in various ways, or organizing it.

By connecting unified silos of enterprise data into project-specific notebooks, which can then be processed by an AI, employees can “synthesize, uncover insights, and enjoy new ways of engaging with data, such as podcast-like audio summaries,” according to Google’s Tiwary.

Let’s wrap this up with a particularly telling statement by Nokia’s Chief Digital Officer, Alan Triggs. He says, “Google Agentspace has the potential to revolutionize how our teams across Nokia find and leverage critical insights.”

He summarized what seems to be the key benefit of this new offering, saying, “We’re particularly excited by Google Agentspace’s ability to blend various data sources and deliver personalized, contextually relevant answers. By unifying our knowledge resources, providing AI-powered assistance and automating workflows, we strive towards reduced time spent searching for information, faster decision-making, and improved collaboration and productivity.”

What do you think about Agentspace? Can your organization benefit from unified silos and agent automation? Have you worked with agentic AI before? Let us know your thoughts in the comments below.

You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, and on YouTube at YouTube.com/DavidGewirtzTV.