Google Agentspace vs Gemini Enterprise: The Rename, Pricing and Connectors

Gemini Enterprise. Google announced the change on October 9, 2025, folding Agentspace into a single enterprise product line. The clearest proof that the change is complete rather than cosmetic is in the URLs: as of August 20, 2026, both cloud.google.com/agentspace and cloud.google.com/products/agentspace return HTTP 301 redirects to cloud.google.com/gemini-enterprise. Google's current product documentation contains no mention of the Agentspace name at all.

Announcing it, Sundar Pichai described the result as "an AI-powered conversational platform designed to bring the full power of Google AI to every employee for every workflow." Thomas Kurian, who runs Google Cloud, framed the agent side as a no-code workbench where any user "can analyze information and orchestrate agents to automate processes."

Agentspace itself was not old. Google Cloud launched it on December 14, 2024 and ran it through most of 2025 on an allowlist rather than open general availability. So the product spent roughly ten months under its original name, which is why so much of the material describing it is both recent enough to look current and old enough to be wrong.

Why this matters more than a normal rebrand

Because a second Google product was renamed into almost the same words six months later. At Cloud Next in April 2026, Vertex AI became the Gemini Enterprise Agent Platform. There are now two live Google products whose names differ by two words, sold to different buyers, billed on completely different models. One is a per-seat application for employees. The other is consumption-billed cloud infrastructure for engineers. "Gemini enterprise vs gemini enterprise agent platform" is a real autocomplete suggestion, which tells you how many people hit this wall.

If you are writing a requirements document, name the product and paste the URL. It is the only reliable way to make sure everyone in the approval chain is discussing the same thing.

Functionally, less than the naming drama suggests. The three things Agentspace was built to do are the three things Gemini Enterprise does now.

The substantive change is the last row. The edition structure was rebuilt, and edition is now what decides whether a given capability is available to you at all. That is the part a pre-October-2025 comparison cannot tell you, and it is where the money is.

This is the single most useful table on this page, and it is the one buyers most often skip. Every value below is read directly off Google's published edition comparison, checked on August 20, 2026. Quoted strings are Google's own wording.

Read the Agent Marketplace column on its own. Business is the only edition in Google's comparison with no check against "Access the Agent Marketplace." It is also the only edition that does not "Access the full data connector ecosystem," and the only one missing "Use Gemini Code Assist Standard," "Use Data Insights (Made by Google)," "Get priority access to the latest Gemini models," "Use your own full-code agents built outside of Gemini Enterprise," and "Enterprise-grade security and compliance."

Business is the edition aimed at organizations of up to 500 users. Those are precisely the organizations least likely to have a team of engineers spare to build agents from scratch, and they are the ones locked out of the catalog of agents other people already built. If installing partner agents is anywhere in your requirements, Standard is your floor, not your upgrade path.

Frontline is an add-on, not a cheap way in

Frontline gets misread as the entry tier because it has the smallest storage allocation. It is not an entry point at all. Google lists it as requiring "150 or more users of Standard or Plus editions" already in place. It ships 2 GiB of pooled storage, cannot build or publish no-code agents, and excludes Code Assist. It does, interestingly, carry both the full connector ecosystem and Agent Marketplace checks, so deskless staff on Frontline can use agents that somebody with a Standard or Plus seat provisioned for them. It extends a deployment. It does not start one.

Agentspace pricing is a dead question, because Agentspace is no longer sold. What you can price today is Gemini Enterprise, and the honest answer is that Google does not publish a complete public price list for it. There is no pricing page with five numbers on it. Every figure circulating online is reseller-reported or comes from someone's quote.

The figures most consistently reported by Google Cloud partners are roughly $21 per seat per month for Business, $30 for Standard on a 12-month commitment, and $50 for Plus on an annual commitment, with higher numbers where there is no commitment. Treat those as a bracket for your own negotiation rather than as published prices, and get the edition written into the quote alongside the number.

Two costs sit outside the seat line and catch people out. Compute consumption bills separately against your linked Google Cloud account rather than being bundled into the seat, so the seat price is a floor rather than a total. And on a platform whose entire value depends on connecting your data sources correctly, deployment and change management are genuine line items, not rounding. Our full breakdown of the numbers, including license cost per seat and the seat minimums, is in the Gemini Enterprise pricing analysis.

Is Google Agentspace free?

No. There is no free tier of Gemini Enterprise and there was no free tier of Agentspace. The cheapest genuinely available way to try the product is the Pay-as-you-go edition, which Google lists as having a "1 minimum seat requirement to get started," so a single seat is enough to open an account rather than a 150-seat floor. Free Gemini tiers that do exist are consumer or Workspace products, not this one, and they do not index your company data.

Connectors are the part that decides whether any of this works, because an agent that cannot reach the system holding the answer is decoration. Google's documentation describes "prebuilt connectors for commonly-used third-party applications like Confluence, Jira, Microsoft SharePoint, and ServiceNow," and the connector section of the documentation runs to well over a hundred individual entries.

Three things are worth checking before you assume coverage.

Your edition. Covered above: Business gets "select, segment-relevant connectors" rather than the full set, and Google does not publish which connectors fall inside "segment-relevant" for a given customer. That is a question for your account team, in writing, with your actual system list attached.

Permission fidelity. Search and agents are permissions-aware, meaning they respect the access controls in the source system rather than flattening everything into one index. This is the property that makes enterprise search safe to switch on across HR and finance data, and it is worth verifying during a pilot with a deliberately restricted test account rather than taking on trust. Set up an account that should not see a specific document, then try to make the assistant surface it.

What is not on the list. The connector catalog is broad and Google-shaped. If the system your agent needs is a line-of-business application, a regional accounting package, or anything you built yourself, you are looking at a custom connector project, which is a development effort rather than a configuration screen. Price that before you sign, not after.

If your organization signed an Agentspace contract in 2025, the practical position reported by Google Cloud partners is that existing deployments kept running unchanged, that Agentspace stopped being available for new subscriptions at the end of 2025, and that contracts move onto Gemini Enterprise at renewal. Google has not published a standalone customer migration notice we can point you at, so treat those dates as partner-reported and confirm your own against your agreement.

What is worth doing regardless of what your paperwork says:

1. Get the target edition in writing. The old tier you bought does not map cleanly onto the current five editions. Ask which one your renewal lands on and get the answer in the quote rather than in an email thread.
2. Re-check the Agent Marketplace and connector rows against that edition. This is where a renewal can quietly remove a capability you are already using. If you land on Business, you lose marketplace access.
3. Inventory what your team built. Custom agents, connector configurations and any indexed data stores are the assets at risk in any migration. Write them down before the conversation, not after.
4. Check who is actually using it. Seats bought and seats used diverge fast on platforms like this, and renewal is the one moment you have leverage. Pulling the usage numbers, and understanding what people actually do with the tool once it is deployed, turns a renewal into a negotiation instead of a rubber stamp.

Google named Banco BV, Cohesity, Gordon Food Services, KPMG, Rubrik and Wells Fargo among early Agentspace adopters. Wells Fargo is the one with published detail, and it is the most useful reference point for any US organization weighing this up, because the work described is document-heavy back-office work rather than a demo.

Google Cloud's own write-up, with Tracy Kerrins, Wells Fargo's Consumer CIO and Head of Enterprise Generative AI, describes three workloads. In corporate and investment banking, the bank is "building agents to help employees answer, triage, and summarize complex foreign exchange post-trade inquiries, as well as navigate policies and procedures across internal data sources and systems." In contract management, the scale is the striking part: Wells Fargo "handles roughly a quarter of a million documents related to vendor agreements," and a custom agent queries them to identify "contracts with specific clauses, payment terms, contract types, and other important contract information." In customer service, agents handle routine work such as "balance inquiries and debit card replacements."

The pattern across all three is the same, and it is the pattern that justifies a platform purchase: a very large corpus of internal documents, strict permissions, and staff who currently answer questions by searching several systems by hand. If your organization does not have that shape of problem at that scale, the platform overhead is buying you capability you will not use.

The direct competitors are the other big-vendor enterprise assistants, and each is anchored to a stack.

We have worked the numbers on the closest two: Microsoft Agent 365 covers how Microsoft governs and bills agents, and Glean pricing covers the vendor-neutral option. For a side-by-side of the catalogs themselves, see AI agent marketplaces.

Three situations where this is the wrong purchase

You are under 500 users and want partner agents. The edition that fits your size is the edition that cannot install them. That is a genuine dead end, not a negotiating position, and it pushes you to Standard pricing for a company Google itself sized as a Business customer.

Your stack is not Google-shaped. The connector list is long but it is a list. If the systems holding your answers are not on it, no edition fixes that, and a custom connector is a software project.

You want an agent, not an agent platform. This is the most common one by a distance. Buying Gemini Enterprise to get one working agent means buying a seat licence, a connector configuration, an indexing project and an approval workflow, in order to reach something you could have deployed directly. If a single job is the goal, browse ready-made AI agents by category, check the integrations and permissions each one needs up front, and deploy it onto the stack you already run.