Claude for business, with Wise Pirates.
Fast, sophisticated and safe: your whole company working with Claude, not a few people experimenting alone. We set up Claude Team or Enterprise, train every role and automate real processes end to end. Model-agnostic, so you get Claude where it wins.
Claude is no longer a chat window. It does the work.
Claude is Anthropic's AI. For a company in 2026 it is three things; most teams only use the first.
A workspace for the whole company
Chat, shared Projects, documents, slides and spreadsheets, and search across Google Drive, Gmail, Slack, Microsoft 365 and GitHub.
Claude Code, included, and not only for developers
Team seats and current Enterprise seats include Claude Code: an agent that reads files, runs scripts and builds small tools, for anyone who works in spreadsheets and documents.
Plus Skills (your procedures, packaged) and connectors on the open MCP standard.
The engine for your own automations
The Claude API, the Agent SDK and Managed Agents run processes end to end, with people approving what matters.
From Anthropic directly, or on Amazon Bedrock, Google Cloud and Microsoft Foundry.
The licence is the easy part. You buy Claude from Anthropic or your cloud; our value is adoption and use cases, not resale. Not sure Claude is the right AI for you? See our decision rule.
Your people already use AI. The question is how.
In most companies, AI arrived bottom up: personal accounts, prompts copied from social media, template packs bought online, connectors installed because a video said so. It feels fast. It is fragile, invisible and, once agents can act, risky.
A widely quoted 2025 MIT report put it at 90% of employees using personal AI tools, against 40% of companies with official subscriptions. The method is debated; the direction is not. Shadow AI is the default, unless someone designs the alternative.
Company data in personal accounts is data outside your control. It becomes one company workspace, with SSO and retention set.
What we mean by fast, sophisticated and safe.
The first shared Skills and a first working automation arrive within the pilot, on your own processes.
Skills, connectors and agents built on your procedures, data and brand, each with an owner, tests and a version.
Approved connectors checked against the OWASP MCP Top 10, retention set to policy, every agent action logged, and a person approving what cannot be undone.
They are not a trade-off. Done properly, safe is what makes it fast.
The right AI for the job: a front door for your people, models for your agents.
Choosing AI is not picking a chat app. It is two decisions: the workspace your people use every day, and the models your agents run on. Nobody pays us to pick Claude: we decide by your data, your own cases and cost.
Gemini Enterprise
Google's governed front door, on Google Workspace or Microsoft 365.
Pick it when- Your people work in Google Workspace or Microsoft 365, and data sits across several systems
- You want one permission-aware search, with agents over it
- Video, multimodal work or very high volume at low unit cost
Claude and other non-Gemini models are used there to build agents (Agent Platform, Model Garden), not as the everyday chat, in the Standard, Plus or Pay-as-you-go editions.
Microsoft 365 Copilot
Microsoft's assistant, for an all-Microsoft estate.
Pick it when- 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
Claude
Anthropic's thinking and building tool.
Pick it when- Agents use many tools over many steps, or the job is agentic coding
- The work is long documents, careful writing and complex instructions
- You want Claude inside Excel, Word and PowerPoint, Slack or a code base
ChatGPT or GPT
OpenAI's assistant and models.
Pick it when- Your people already live in ChatGPT
- Image and voice generation are core to the job
- Bulk tasks need very cheap small models
Or a mix. Most of our automations route by task, with a fallback provider and a cost ceiling. Pick the AI that fits, not the one that pays us.
When we would not pick Claude.
- All data must stay in the EU, on Anthropic's own apps. Not possible today: we route through EU clouds or go sovereign.
- You need one governed search across all your systems. Gemini Enterprise or Copilot may be the better front door; our Google Cloud team does Gemini too.
- Images and video are the core of the job. Claude has no native image generation; our AI Factory does.
When you do not need us.
- You only want seats. Buy them directly. We add value on adoption, not a reseller margin.
- 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, publishing or deleting, we always build an approval step.
Four services. Enablement first, then execution.
Most of the value sits in the first two: people using Claude well, every day. Then automation, then the foundation your lawyers and CISO sign off.
We do not sell prompt packs, template bundles or connectors off a shelf. We build yours, on your work, and hand them over.
Adoption: Claude Team and Enterprise, set up properly
From the plan decision to every team using it daily.
- Plan and seat mix decided before you sign
- SSO, SCIM, roles, retention, connectors and spend limits configured
- An acceptable-use policy people actually read
- Training by role, a champions network and a shared Skills library
- A monthly adoption view and a day-90 keep, expand or cut decision
Claude Code and Claude Design for business teams
Builders in marketing, operations and finance, working safely.
- Internal tools, calculators and small apps
- Data analysis and dashboards on your own exports
- Marketing automations: ad variants, feeds, SEO checks
- Prototypes and decks with Claude Design (still a research preview)
Automation and agents, end to end
We design, build and run automations on the best Anthropic model for each step.
- Skills, MCP connectors, the Agent SDK and Managed Agents
- Integrated with n8n, Make, Zapier, your CRM and your workspace
- People approve every irreversible step
- Evals before launch, cost per task under control
- Results: productivity, better customer and employee experience
Governance, security and EU data
The Claude your legal team approves, with guardrails from day one.
- EU AI Act and GDPR: roles, literacy records, transparency notices
- EU data architecture through Amazon Bedrock or Google Cloud
- Wise Shield, our LLM and agent firewall, between agents and your tools
- Agent observability, audit trails and spend alerts
Where Claude pays, team by team.
Start from a real bottleneck, not a licence. These are the patterns we build most, marketing first.
Claude for Marketing, with Wise Pirates.
A brand-voice Skill, a campaign and content workflow with a review step, and a weekly report automated from your ad and analytics exports. The team that sets up Claude also knows the media.
Talk about Claude 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: the brand-voice Skill, the campaign workflow and the weekly report. | Faster production, a report on Monday |
| Finance | The close lives in someone's head and a week of spreadsheets. | Close checklists as Skills, variance commentary in Excel, and board packs drafted in PowerPoint on your template. | A shorter close, fewer errors |
| Customer service | Agents spend the first minutes of every ticket working out what it is about. | An automation that classifies, summarises and drafts replies from your knowledge base, with the agent approving. | Faster resolution, better customer experience |
| Sales | Call prep and proposals eat selling time. | Account research and call prep from your CRM; proposal first drafts on your template. | More time selling |
| HR | The same questions, answered by hand. | Job ads, interview kits, onboarding and policy answers. Never candidate ranking: under the AI Act that is high-risk. | Better employee experience |
Anthropic publishes large time savings from enterprise customers. They are vendor-reported, so we never promise them: we measure yours.
Builders who do not write code.
The surprise of 2026: one of the most useful agents for a marketing or finance team is a coding agent. Claude Code reads and writes files, runs scripts and calls tools, from a desktop app or the browser.
From a spreadsheet problem to a working tool, in an afternoon.
Hands-on sessions on your own files.
- A CSV of ads in, underperformers flagged, new variants out
- A CRM data-quality check that names the owner of every stale deal
- A weekly report pulled from APIs and delivered to Slack as a PDF
On-brand prototypes, decks and one-pagers.
Visual work through conversation, on your design system. Still a research preview, and we treat it as one.
- Your brand system set up and locked, so every output starts on brand
- Clickable prototypes to test an idea before briefing developers
- Sales one-pagers and decks on the corporate template
Our rule for citizen builds: nothing important stays a prototype.
- Step 1PrototypeAnyone can build, on test data, in a sandbox.
- Step 2ReviewedSecurity, secrets and data handling checked by our specialists.
- Step 3OwnedA named owner, a version and a place in your stack.
Anything that touches customer data or becomes a dependency goes through all three. When a tool needs a real backend, we move it to a proper stack, such as Lovable with Supabase.
We do not train people on Claude. We train them on their work, with Claude.
Anthropic's 2026 research found that experienced users get more out of Claude over time. Working well with AI is learned, and it compounds.
- 1EveryoneLiteracy firstWhat Claude is, what never goes in, how to check its work.
- 2Power usersWork, by functionProjects, Skills, connectors and research, on the team's own tasks.
- 3BuildersTools and automationsClaude Code, automations and the API, reviewed before anything ships.
- 4ChampionsOne per teamAbout one per 15 to 25 people, with protected time. They own their function's Skills and are the first line of support.
Know-how that stays when people leave
- Every Skill has an owner, a version, test cases and a changelog
- Published through a private marketplace, reviewed before release
- Retired when stale, not forgotten
From first session to proven
- Anthropic's free courses mapped to each track
- Our own practical exercises, on your processes
- Records you can show under the AI Act literacy duty
pt-PT, not pt-BR by accident
- Delivery and training in European Portuguese, or English
- Anthropic publishes no European Portuguese benchmark, so we ship a pt-PT style Skill
- A reviewer step on anything customer-facing
How we measure it.
Training runs with our AI Training & Upskilling team, from a half-day leadership session to a full academy by role.
We run Claude ourselves. Every day.
We do not teach from slides. What we set up for you, we already run for ourselves, built by the same Claude specialists (Skills, connectors, Claude Code and the API) who will work with you.
What runs inside Wise Pirates.
Built in Portugal, delivered in Portuguese and English, with GDPR, CNPD guidance and the AI Act mapped for Portuguese teams.
A small first step. A written decision at day 90.
Two-week sprints, a weekly demo, every decision in writing. Start at any step.
- Step 1Working sessionHalf a day with leadership and future champions: what Claude can do for you, and where to start.You leave with a shortlist of use cases.
- Step 2Discovery sprint2 to 3 weeks, fixed scope. Current AI use, shadow AI included; data classes; use cases scored on value, feasibility and risk; a baseline.You leave with a ranked roadmap.
- Step 3Pilot4 to 8 weeks. 20 to 50 users in two or three functions, SSO on, first Skills, one or two automations with evals.You leave with results on real work.
- Step 4RolloutFunction by function: approved Skills and connectors, training by role, the monthly adoption view.You leave with company-wide adoption.
- Step 5Centre of competenceOwns Skills, connector approvals, evals, model upgrades and costs, with us or on its own.You leave with a capability, not a dependency.
Forty Skills. One context window. No clutter.
A Skill is a folder: instructions, templates and scripts for one job. Claude keeps only each Skill's name and a one-line description in view, and loads the rest when a request needs it. So a company can install dozens of Skills without crowding Claude's working memory, its context window.
Done, with 39 Skills never loaded.
The right model per step. A person at the point of no return.
A pattern we build: invoices arriving by email, processed without retyping. A workflow tool owns the trigger and the retries, Claude does the judgement, and a person approves before anything is posted. Every step leaves a line in the trail, and every document has a price.
$0.0405 for a flagged invoice. Flags are the exception, not the rule.
Built and run by our Process Automation team, with every run traced by Agent Observability.
Our go-live checklist. Six lines, all ticked.
Nothing touches real company data until every line is ticked. Done once, properly, it is reused for every new Skill and automation.
- 01Company accounts only, SSO on, personal accounts retired
- 02Retention, connectors and spend limits set to policy
- 03Every Skill and connector approved, owned and versioned
- 04Evals passed and the model version pinned
- 05Human approval on every irreversible step; logs flowing to your SIEM
- 06AI Act literacy records and transparency notices in place
Need the data to stay in Europe? There is a route.
Anthropic's own apps and API process data globally or in the US, with no first-party EU data residency as of September 2026. For workloads that must stay in the EU, we architect Claude through the clouds that offer it.
Microsoft Foundry has an EU data zone announced for 2026; we check it per project. For stricter needs, our Sovereignty area goes further, and cloud architecture runs with our Cloud Services and Google Cloud teams.
Ten questions your CTO should ask. Our answers.
If your CTO reads one section, make it this one.
01Identity and access
Company accounts only, provisioned from your identity provider.
- SSO with SAML or OIDC, domain capture, SCIM provisioning from Okta or Entra
- Roles, custom roles and group spend limits
SCIM, audit logs, IP allowlisting and customer-managed keys are Enterprise features.
04Connectors and MCP
Every connector is a permission you grant.
- An approved list; vendor-official remote servers with OAuth; versions pinned
- Tool descriptions reviewed, tokens scoped per user, every tool call logged
Mapped to the OWASP MCP Top 10, from tool poisoning to supply chain.
06Firewall and observability
A control point between agents and your systems.
- Wise Shield inspects prompts, tool calls and outputs against your policy, with Agentic Security
- Traces per run: tokens, tools, latency, cost and eval score, with Agent Observability
02Data handling
No training on your content, by contract.
- The Commercial Terms bar training on customer content
- Retention set to your policy; zero data retention by agreement on the API, not the apps
03Audit and compliance
Everything that happens is exportable.
- Audit logs and the Compliance API for eDiscovery and DLP
- OpenTelemetry from Claude Code and the desktop agent into Splunk, Datadog or Elastic
05Browser and computer use
Powerful, and the easiest to attack.
- Claude in Chrome, on the paid plans since August 2026, restricted to approved domains
- A separate browser profile for agents; approval before any send, pay, publish or delete
07Evals and model upgrades
No silent regressions.
- A golden set of 30 to 200 real cases per workflow, with sampled human review
- Versions pinned in production; upgrades pass the evals first
08Agent runtime and capacity
The right harness for each job.
- Workflow tools own triggers and integrations; the Agent SDK or Managed Agents run the agents
- Rate-limit tiers, retries with backoff, a fallback provider, step and budget limits
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 or credit flagged (from 2 December 2027); a DPA, a transfer mechanism and a DPIA where needed
10Portability and exit
You can leave Claude, and you can leave us.
- Skills follow an open standard, tools use MCP, and evals are provider-neutral
- Everything we build lives in your accounts, your repositories and your cloud
Switching model is a test run against your golden set, not a rebuild.
Model note, 1 October 2026: the current line-up is Claude Fable 5.1 (1 September), Opus 5.5, Sonnet 5.5 (28 September) and Haiku 4.5. Fable and Mythos models retain data for 30 days and are not eligible for zero data retention by default. Names change; our routing and evals do not depend on them.
Claude is the start. Put it to work.
These teams sit right next to Claude adoption.
On a new project, the first step is an assessment: our discovery sprint.
Assessment is one of our specialties: how your teams use AI today, so every decision rests on what is really there. A working session is half a day. Talk to our specialists to see how this applies to your case.
The discovery sprint
Current AI use mapped (shadow AI included), data classes, processes worth automating and your AI Act roles, use cases scored on value and risk, a baseline and a ranked roadmap you keep.
The seat review
We find the Team or Enterprise seats nobody uses and turn them into workflows, or tell you to cut them. We do not sell licences, so cutting costs us nothing.
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.
Claude for business, answered.
Is "Claude for business" an Anthropic product?
Claude Team or Claude Enterprise: which plan does our company need?
How much does Claude cost for a business?
Does Anthropic train Claude on our company data?
Can Claude keep company data in the EU?
What does the EU AI Act mean for a company using Claude?
Can we use Claude inside the Microsoft 365 or Google Workspace we already have?
Is Claude Enterprise GDPR compliant?
How do we stop employees using personal Claude accounts for work?
Are Claude MCP connectors safe for company data?
Claude vs ChatGPT vs Gemini Enterprise vs Microsoft 365 Copilot: which is best for business?
Who implements Claude for companies in Portugal?
How long does it take to roll out Claude in a company?
Can non-developers use Claude Code?
Can Claude automate a business process end to end?
Fast, sophisticated and safe. All three.
Start with a working session or a discovery sprint: results on real work early, decisions in writing, everything in your own accounts.
Claude and Anthropic are trademarks of Anthropic, PBC. The Claude logo is shown for identification only. Wise Pirates is an independent agency and is not affiliated with or endorsed by Anthropic. Other product names are trademarks of their owners.
