Gemini Enterprise, licensed and put to work.
Your files, your permissions, one Google partner: the whole company working with Gemini Enterprise, not a few people experimenting alone. We resell the licences as a Google Cloud partner, connect it to Workspace or Microsoft 365, train every role and automate real processes end to end: fast, sophisticated and safe. Model-agnostic, so you get Gemini where it wins.
Start with a working session or a 2 to 3 week discovery sprint.
One front door to AI for every employee. A platform to build behind it.
Gemini Enterprise is Google's AI for companies. Since April 2026 the name covers two products: the app your people use, and the Agent Platform, formerly Vertex AI, where agents are built.
It is not the Gemini inside Gmail and Docs, which comes with Workspace. Gemini Enterprise is a separate subscription that works with Google Workspace or Microsoft 365, and does not need Workspace.
One assistant that knows your company
Chat, search and research across Drive, Gmail, SharePoint, Outlook, Salesforce, Jira and more, each person seeing only what they may open.
Plus Deep Research, Gemini Notebook, Canvas and shared Projects.
Teams build agents without code
Workflow Builder turns a repeated task into an agent that runs on a schedule, a trigger or request. Skills package your procedures.
Google, partner and your own agents share one gallery and one set of controls.
The engine for your own automations
Agent Studio, the open-source ADK and Agent Runtime build agents wired into your systems.
Model Garden offers 200+ models, Claude included, behind the same governance.
Building and registering custom agents in the app (Agent Studio, ADK, Agent Runtime, and non-Gemini models such as Claude from Model Garden) needs the Standard, Plus or Pay-as-you-go editions. The Agent Platform itself is a Google Cloud service, billed by consumption.
The whole Google stack, one team.
Premier Partner in Ads. Cloud and Workspace partner.
Reseller and service partner for Google Cloud and Workspace. Nine years with Google, 300+ brands on Google ecosystems.
We run the marketing, not only the AI
Strategy, media, content and measurement, every day. Our agents automate work we already do well.
One team across Ads, Cloud, Workspace and Gemini
They know the Workspace or Microsoft 365 it searches and the Ads and BigQuery data it reports on. One brief, no hand-offs.
Gemini on the data we already run
Already running your Google Ads, GA4 and BigQuery? They become your agents' first sources, with no data project first.
See Gemini for MarketingOn Microsoft 365? It works for you too: it connects to SharePoint, OneDrive, Outlook and Teams.
Seven habits that feel fast. Each gets a governed twin.
In most companies AI arrived bottom up: personal accounts, copied prompts, weekend agents, connectors switched on because a video said so. Once agents can act, it is risky.
Gemini Enterprise adds a risk of its own: every folder shared with the whole company years ago becomes an answer today.
Same people, same speed. Now it is governed, shared, capped and measurable.
A front door for your people. The right models for your agents.
Not a pick of chat apps: two choices. The platform your people work in every day, and the models your agents run on. We decide both by where your data lives, results on your cases and cost.
Pick Gemini when
Gemini Enterprise: the governed front door.
- Your people work in Google Workspace or Microsoft 365, and data sits across several systems
- You need one permission-aware search and governed agents over all of it
- BigQuery, Looker, video and multimodal work belong behind one control plane
Pick Copilot when
Microsoft 365 Copilot: an all-Microsoft estate.
- Files and identity live in Microsoft 365 and Entra
- IT wants a single vendor and in-tenant controls
- Your agents will live in Teams and Copilot Studio
Pick Claude when
Anthropic's Claude: the thinking and building tool.
- The work is long documents and careful writing
- Agents use many tools over many steps, or it is agentic coding
- Power users need depth more than search
Pick GPT when
OpenAI's ChatGPT: where it may fit better.
- Your people already live in ChatGPT
- Image and voice generation are core to the job
- Your developers already build on OpenAI's platform
Inside Gemini Enterprise, other models build agents. Claude and other non-Gemini models are used to build agents on the Agent Platform, through Model Garden, not as the everyday chat; that needs the Standard, Plus or Pay-as-you-go editions. Pick the AI that fits, not the one that pays us.
When we would not pick Gemini Enterprise.
- The newest models must be processed in the EU. Today only one Flash model is. We check the model list or go sovereign.
- Everything lives in Microsoft and IT wants one vendor. Copilot is the shorter path.
- A few experts need a writing and coding tool. A handful of Claude or ChatGPT seats may do more, for less.
When you do not need us.
- A small team, a few seats. The Business edition is self-serve. We can bill it, but you may not need a project.
- A one-off prompt problem. You need a short training session, not a programme.
- You want full autonomy, no human in the loop. For sending, paying, granting access or deleting, we always build an approval step.
The right edition, chosen on your real use.
Five ways to buy, by what they include. Google publishes list prices on its Gemini Enterprise page; we confirm scope and price before you sign.
Google keeps evolving its services and editions; we help you choose the best option for each moment. Our rule of thumb: seats for the broad workforce, consumption for the 5 to 10% of workflows that run as heavy agents.
Business
Small teams, starting fastNo IT setup, self-serve, a 30-day trial- The assistant, search and content creation
- Connectors, Microsoft 365 and Workspace included
- No-code agents; Google-made agents such as Gemini Notebook
Not included: the full connector range, code or third-party agents, Data Insights, enterprise controls.
Standard
Companies that need IT controlsUnlimited seats, a 30-day trial- Everything in Business, higher quotas
- Your ADK and third-party agents, Marketplace, Data Insights
- VPC Service Controls, customer-managed keys, data residency
Plus
Heavier daily useThe Standard controls, more headroom- Everything in Standard
- Higher daily quotas, for Deep Research and agent creation
- More pooled storage per seat
Pay-as-you-go
Agent workloads and uneven useNo seat fee, generally available since 1 August 2026- Core app capabilities, with no seat fee
- Usage billed at Agent Platform rates for tokens, compute and storage
- Needs an invoiced Cloud billing account
Gemini Notebook is not included yet.
Frontline
Workers who use, not buildAn add-on to Standard or Plus- Shop floor, stores, field teams
- Use agents and search; cannot create agents
- Requires 150 or more Standard or Plus users
Cost per run, as an index. The levers we pull.
One agent workflow on the Agent Platform: a weekly report from your Ads, GA4 and BigQuery data, approved by a person. The baseline run is 100. Seats include pooled daily quotas; the agent itself is billed by consumption either way.
Caching and batch together: half the baseline run, for the same report. Now try Gemini 4 Argon.
Bought through Wise Pirates: one partner, one invoice.
From licence to daily habit to agents that do the work.
Most of the value sits in the first two: people using Gemini Enterprise well, every day. Then automation, then the foundation your CISO signs off.
No prompt packs, template bundles or agents off a shelf. We build yours, on your work, and hand them over.
Licences and setup, done properly
From the edition decision to daily use.
- Edition and seat mix decided before you sign
- Identity, location, licences, quotas and spend caps configured
2 more
- Sharing cleaned and sources scoped before anything is indexed
- Training by role, a champions network and a shared Skills library
Agents for business teams, without code
Marketing, sales, finance and operations.
- Workflow Builder agents on schedules, triggers and chat
- Skills for your brand voice, report formats and procedures
2 more
- Deep Research and Data Insights on your sources
- Always reviewed, always with an owner
Automation and agents, end to end
Agents wired into your own systems.
- The Agent Platform: Agent Studio, ADK, Agent Runtime, Memory Bank
- Connected to your ERP and CRM over MCP, to agents over A2A
2 more
- People approve every irreversible step
- Evals before launch, cost per run under control
Governance, security and EU data
Guardrails your legal team approves, from day one.
- EU AI Act and GDPR: roles, literacy records, transparency notices
- EU location and keys, gaps mapped per feature
3 more
- Model Armor and Agent Gateway set up; Wise Shield across every AI
- Agent observability, audit trails and spend alerts
- Portability: agents, Skills and evals that live in your accounts
Eight bottlenecks it clears, on data you already have.
Start from a real bottleneck, not a licence. These are the patterns we build most, marketing first.
Results Google reports are vendor-reported: we never promise them, we measure yours against your baseline.
Gemini for Marketing.
Three agents on your own brand and data: a brand-voice Skill, a campaign agent with a review step, and a reporting agent on your Google Ads, GA4 and BigQuery data.
Talk about Gemini for Marketing →| Team | The bottleneck | What we build | Result |
|---|---|---|---|
| Marketing | Dozens of variants per campaign, and a weekly report that eats a day. | The offer above, on your own brand and data. | Faster production, a report on MondayGoogle reports Virgin Voyages uses 50+ agents to cut campaign creation time by 40%. |
| Sales | Call prep means five tabs: CRM, mail, proposals, notes, news. | An account brief before every meeting, from CRM, mail, proposals and the web. | More time selling |
| Customer service | Answers hide in policies, contracts and old tickets. | An answer agent for your people over policies and contracts, permission-aware, with citations. | Faster, consistent answersGoogle reports SIGNAL IDUNA delivers insurance answers 37% faster. |
Show all 8 teams
| Finance | Month-end commentary written by hand. | Variance commentary drafted from your data and the threads behind it, with Data Insights. | A shorter close, fewer errors |
|---|---|---|---|
| Operations | Knowledge split across SharePoint, Jira and an ERP. | One search across all three, plus scheduled agents for recurring reports. | Hours back, every week |
| HR | The same questions, answered by hand. | A policy concierge and an onboarding agent. Never candidate ranking: under the AI Act it is high-risk. | Better employee experience |
| Legal | Every contract waits in the same queue. | Contracts checked against your playbook, with Google's legal edition (preview) where it fits. | Faster turnaround, same standards |
| Leadership | The board pack is a week of chasing numbers. | A scheduled weekly brief: KPIs, top threads and risks, each with its source. | Better-informed decisions |
Your people learn it on their own files, in their own language.
Training is part of every rollout, by role, on each team's own tasks.
- 1EveryoneLiteracy firstWhat it is, what never goes in, how to check its work and sources.
- 2Power usersWork, by functionSearch, Deep Research, Notebook and Skills, on the team's own tasks.
- 3BuildersAgents and automationsWorkflow Builder, then Agent Studio and ADK, reviewed before shipping.
- 4ChampionsOne per teamWith protected time, they own their function's Skills and agents.
Know-how that stays when people leave
- Each with an owner, a version and test cases
- Shared through the agent gallery, reviewed before release
- Retired when stale, not forgotten
pt-PT, not pt-BR by accident
- Google publishes no pt-PT quality benchmark for the app
- So we ship a pt-PT glossary Skill and a reviewer step
- Voice agents tested in European Portuguese first
How we measure it.
Training runs with our AI Training & Upskilling team, from a half-day session to a full academy, with AI Act literacy records.
Discovery, pilot, rollout. A written decision at day 90.
Two-week sprints, a weekly demo, every decision in writing. Start at any step.
- Step 1Working sessionYou leave with a shortlist of use cases.
What happens
Half a day with leadership and future champions: where to start. - Step 2Discovery sprintYou leave with a ranked roadmap.
What happens
2 to 3 weeks, fixed scope: our assessment. What is in it. - Step 3PilotYou leave with results on real work.
What happens
4 to 8 weeks, on Google's 30-day trial where it applies. 20 to 50 users, first Skills and agents, one automation with evals. - Step 4RolloutYou leave with company-wide adoption.
What happens
Function by function: licences assigned, sources added in order, training by role, the monthly adoption view. - Step 5Centre of competenceYou leave with a capability, not a dependency.
What happens
Owns Skills, agents, approvals, evals and costs, with us or on its own.
On a new project, the first step is an assessment: our discovery sprint.
Assessment is one of our specialties. Talk to our specialists to see how this applies to your case.
The discovery sprint
Your edition and seat mix, current AI use mapped (shadow AI included), sharing reviewed on the first sources, use cases scored on value and risk, a baseline and a ranked roadmap you keep.
The licence rescue
Adoption stalled? We find out why, from usage data and the people who stopped, then fix the top blockers: sources, sharing, training or the wrong edition.
One question. Every system. Only what you may see.
A question goes to every connected source at once. Each source applies its own permissions, so the answer uses only what the person asking may already open, with citations.
One question, asked once.
Agents that delegate. A gateway on every hop. A person at the point of no return.
A pattern we build: a manager asks for a new hire to be onboarded. The assistant hands each part to the agent that owns it over A2A, the open agent-to-agent protocol; agents reach their tools over MCP.
In our setup (EU or US location), every hop passes Agent Gateway: identity checked, Model Armor screening, unapproved calls denied. Sensitive access waits for a person.
The HR agent writes the record through its tool, over MCP.
Six switches we turn on before your data goes in.
No real data goes in until every switch is on. Set once, reused for every new source and agent.
- 01Company tenant only, SSO on, location chosen on purpose
- 02Sharing reviewed and sensitive sources excluded before indexing
- 03Every connector, MCP server and agent approved, owned and versioned
- 04Evals passed on real questions, permission cases included
- 05Human approval on every irreversible step; quotas and spend caps set
- 06AI Act literacy records and transparency notices in place
6 of 6 on· ready for your data
Need the data to stay in Europe? Know where the edges are.
Gemini Enterprise has a real EU option, and real gaps. Google itself recommends the global location unless you have a regulatory reason.
For regulated sectors we design per feature: EU location, customer-managed keys, Web Grounding for Enterprise (EU-resident, unlike Grounding with Google Search), Model Armor and a pinned model list. For defence and critical infrastructure, our Sovereignty area goes further, with our Cloud Services and Google Cloud teams.
Ten questions your CTO should ask. Our answers.
01Permissions and indexing
Answers follow permissions. So does over-sharing.
- Federated connectors query live; ingested ones copy content and access lists into a Google-managed index
- Sharing reviewed and sensitive sources excluded before the first index
02Connectors, MCP and A2A
Every connection is a permission you grant.
- Custom MCP off until an organisation policy allows it; A2A agents registered one by one, with OAuth
- Agent Registry as the catalogue; Agent Gateway allow and deny rules on every outbound call
03Firewall and data protection
A control point between agents and your data.
- Model Armor screens for injection, jailbreaks and data leakage, in the EU and US locations, not in global
- Sensitive Data Protection masks personal data; Wise Shield applies one policy across every AI you run, with Agentic Security
Show 7 more questions
04Identity and access
Your identity provider stays in charge.
- Google identity, or Workforce Identity Federation with Entra ID, Okta or any SAML or OIDC provider
- Attribute mapping must match each source's user IDs; each agent gets its own Agent Identity, separate from users, so tool access is least-privilege
05Location, keys and perimeter
The location is chosen once, when the app is created.
- Global, EU or US; customer-managed keys only in the EU and US
- VPC Service Controls around the app; Access Transparency on Standard and Plus
With Grounding with Google Search on, residency and perimeter guarantees do not apply.
06Audit and observability
Every run leaves a trace.
- Cloud Audit Logs for admin and data access; Model Armor verdicts logged
- Traces, latency and error rates per agent, with Agent Observability
07Evals and model upgrades
No silent regressions.
- Agent Simulation before launch, Agent Evaluation on live traffic, Agent Optimizer clustering failures into new golden questions: 50 to 200 per agent
- Re-run after every model change: Google shipped four Flash versions between May and September 2026
08Cost control
Seats for the many, consumption for the heavy.
- Pooled daily quotas; when they run out, stop or allow overage at pay-as-you-go rates
- Monthly spend caps and alerts; agent runs modelled separately
Cost per run is modelled before we build; see the index under Editions.
09EU AI Act and GDPR
Your duties as a deployer, mapped.
- AI literacy (Article 4) with training records; transparency duties (Article 50) from 2 August 2026
- High-risk uses such as recruitment flagged (obligations from 2 December 2027); a DPIA where needed
10Portability and exit
You can change model, and you can leave us.
- ADK is open source, A2A and MCP are open protocols, and 200+ models run in Model Garden
- Everything we build lives in your Google Cloud project, your repositories and your accounts
Switching model is a test run, not a rebuild.
Product note, 1 October 2026: the app runs Gemini models only, default Gemini 3.8 Flash. Gemini 4 Argon, Google's new frontier model, was announced on 30 September 2026 and is rolling out, first to trusted cyber defenders, then to paid API customers. In the EU multi-region only Gemini 3.5 Flash has EU processing; 3.6, 3.7 and 3.8 Flash and Gemini 3.1 Pro (limited availability) are global only. Renames: Agentspace is the Gemini Enterprise app, Vertex AI the Agent Platform, Agent Designer Workflow Builder, NotebookLM Enterprise Gemini Notebook Enterprise. Names change; our evals do not.
Gemini Enterprise is the start. Put it to work.
Workspace is not a prerequisite: it is one of the systems we connect, next to Microsoft 365 and the rest of your stack.
Your business result. A great team and five pillars to move it.
Two pillars we run for you, three we own, and the crew at the centre answering for your number.
Judged on your KPIs, with part of our fee on the line.
Gemini Enterprise, answered.
What is Gemini Enterprise?
Is Gemini Enterprise the same as Gemini in Google Workspace?
Can we use Claude or other non-Google models in Gemini Enterprise?
How much does Gemini Enterprise cost, and which edition do we need?
Can we buy Gemini Enterprise from a Google partner in Portugal?
Does Google train Gemini Enterprise on our company data?
Does Gemini Enterprise support EU data residency?
Do we need Google Workspace to use Gemini Enterprise?
Can Gemini Enterprise show employees files they should not see?
Is it safe to connect agents and MCP servers to Gemini Enterprise?
Gemini Enterprise vs Microsoft 365 Copilot vs ChatGPT vs Claude: which is best for business?
Who implements Gemini Enterprise in Portugal?
How long does a Gemini Enterprise rollout take?
Can Gemini Enterprise work with our Google Ads, GA4 and BigQuery data?
Can Gemini Enterprise automate a business process end to end?
Fast, sophisticated and safe. All three.
Safe is what makes it fast: once sharing, connectors and approvals are set, every new team starts in days, not months. Start with a working session or a discovery sprint, licences on one invoice, everything in your own accounts.
Google, Gemini, Gemini Enterprise, Google Workspace and Google Cloud are trademarks of Google LLC. The Gemini logo is shown for identification only. This page is published by Wise Pirates, an independent agency, not by Google. Other product names are trademarks of their owners.
