AI Agents & Assistants
Claude for Work
Rest of world Report an error0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.
by Anthropic PBC · claude.com
Compare with Langdock → Report an error on this page Is this your product? →
Link to this page
Add ?theme=dark to the image address for dark backgrounds.
The badge shows the category and the month of the latest evaluation, and EU-Made where
that is the product’s provenance. It never shows a score.
Claude for Work at a glance
- Vendor
- Anthropic PBC
- Headquarters
- Not stated on the vendor’s pages as of 26 Oct 2026
- Data hosting
- Not stated on the vendor’s pages as of 26 Oct 2026
- Free plan
- Yes, per the vendor’s own page source, captured 26 Oct 2026
- Panel verdict
- The AI panel rates Claude for Work 3/5, the mean of 6 judges. Claude for Work — Anthropic's Team and Enterprise plans — earns its highest marks on data use and training and admin and access: the commercial terms state Anthropic may not train models on Customer Content, accepting those terms also accepts the data processing agreement, and Enterprise plans carry single sign-on with domain capture, SCIM provisioning, custom roles with per-connector permission levels, and retention controls with every change tracked in audit logs. how scores are made
- As of
- Vendor pages captured 26 Oct 2026; evaluated by the panel 28 Oct 2026. all sources
Read this page as one judge. Each weighs the same scores by what they care about.
The panel's verdict
Claude for Work — Anthropic's Team and Enterprise plans — earns its highest marks on data use and training and admin and access: the commercial terms state Anthropic may not train models on Customer Content, accepting those terms also accepts the data processing agreement, and Enterprise plans carry single sign-on with domain capture, SCIM provisioning, custom roles with per-connector permission levels, and retention controls with every change tracked in audit logs. The lowest marks go to processing and models and sovereignty: the privacy policy states personal data is transferred to servers in the US or other countries outside the EEA and UK, EEA, Swiss and UK customers contract with Anthropic Ireland Limited under standard contractual clauses, and we found no public information on EU hosting or EU inference. Agent capabilities scored a flat 7 — candid about run limits, yet the same pages state there is no approval step before a run starts. The judges did not genuinely split: no category spans more than two points and the flagged splits came back empty.
Speaks for it
- The commercial terms state Anthropic may not train models on Customer Content, an opt-out is shown on all plans, and accepting the Commercial Terms also accepts the data processing agreement
- Enterprise plans carry single sign-on with domain capture, SCIM provisioning, audit logs, usage analytics, and custom roles that set model access, connector permission levels and per-group spend limits
- Enterprise admins can set custom retention for conversations, projects and artifacts from a 30-day minimum, with permanent deletion and every retention action tracked in audit logs
- Managed-agent documentation states its limits openly — up to 1,000 agents per run, 64 concurrent threads, a 24-hour default run lifetime — and newly created roles default to "Needs approval" on every connector
- Seat prices are published for the Free, Pro, Max and Enterprise tiers, with Enterprise at $20 per seat per month plus usage billed at API rates, a 20-seat minimum and an annual commitment
Held against it
- Data is retained indefinitely by default, custom retention is Enterprise-only, and the controls exclude Claude Managed Agents and features running Claude Code in the cloud
- The privacy policy states personal data is transferred to servers in the US or other countries outside the EEA and UK, and we found no public information on EU hosting or EU inference for the service
- The managed-agents documentation states there is no approval step before a run starts, and we found no public information on test or evaluation tooling before an agent is released
- Usage is billed at API rates specified on a separate Model Pricing Page, usage limits are not quantified, and we found no public information on a Team seat price, so a full annual cost cannot be computed from these pages
- Training use continues for safety-flagged conversations and user-reported materials even after an opt-out
Best for
- You need contractual no-training terms and auditable retention for a works-council or data-protection agreement, with SOC 2 and ISO 27001 stated
- You are rolling the assistant out across many seats and need single sign-on, SCIM provisioning and per-role control over models, connectors and admin areas
- Your team works in Gmail, Google Drive, Slack, Microsoft 365 and Pipedrive and wants a CRM connector that respects existing user permissions
Avoid if
- You must keep processing and storage inside the EU — the privacy policy states transfers to servers in the US or countries outside the EEA and UK, and we found no public information on EU hosting or inference
- You need a human approval gate before agent runs start — the documentation states there is no approval step before a run starts
- Your retention policy must cover agent features — the documented controls exclude Claude Managed Agents and features running Claude Code in the cloud, and default to indefinite retention
- You need a fixed, predictable total cost — Enterprise adds usage billed at API rates on top of the seat price, with usage limits not quantified
The scores
| Criterion | Panel median |
|---|---|
| Data use, training & retention | 4 of 5 |
| Models & processing location | 2 of 5 |
| Admin, identity & access controls | 4 of 5 |
| Integrations, connectors & knowledge | 3.5 of 5 |
| Agent capabilities & guardrails | 3.5 of 5 |
| European sovereignty | 2 of 5 |
| Pricing transparency | 3 of 5 |
Data use, training & retention
Show reasoningHide reasoning
How this is scored
What happens to prompts, uploaded files, connected data and outputs: whether any of it trains a model (the vendor's or a model provider's), how long it is kept and where logs live, whether zero-retention is available, and whether a data processing agreement covers it.
0 — No statement on training or retention; the privacy policy is a website policy that does not cover the product.
3 — A general statement that business data is "not used for training", with no word on retention, on the model providers behind the product or on logs.
5 — Business data excluded from training by contract, retention periods stated for chats and files, and a DPA offered to business customers.
8 — No-training commitments stated for the vendor and every model provider it routes to, retention configurable by the admin (including deletion of chat history), zero-data-retention terms with the model providers named, and abuse-monitoring retention described.
10 — Data use fully accounted for: training, retention, logging and abuse monitoring stated per model provider, customer-controlled retention and deletion with an audit trail, outputs and connected data treated as customer data in the DPA, and an independent audit (SOC 2, ISO 27001 or ISO 42001) covering the product.
The Head of IT
The commercial terms state that Anthropic may not train models on Customer Content from the Services, an opt-out is shown on all plans, and accepting the Commercial Terms also accepts the DPA. Enterprise admins can set custom retention for conversations, projects and artifacts from a 30-day minimum, data past the period is permanently deleted, every retention action is tracked in audit logs, abuse-monitoring retention for safety-flagged conversations is described openly, and SOC 2 and ISO 27001 are listed for the Enterprise plan. I hold it just under the top level because retention defaults to indefinite, the controls are Enterprise-only, and they exclude Claude Managed Agents and features running Claude Code in the cloud — exactly what an agent rollout would lean on. 4 7 6 2 3
The Data Protection Officer
The commercial terms state Anthropic may not train models on Customer Content, the DPA is accepted together with those terms, and Enterprise admins get configurable retention for chats, projects and artifacts with permanent deletion and every retention action tracked in the audit log, alongside stated SOC 2 and ISO 27001. However, default retention is indefinite, custom retention is Enterprise-only, and the controls exclude Claude Managed Agents and features that run Claude Code in the cloud — for a buyer asking where a prompt inside an agent run goes, that gap matters. I found no public information on zero-data-retention terms as a general option, though the pages note healthcare customers can sign a BAA with no Zero Data Retention requirement. 4 6 7 2 5
The Automation Lead
The commercial terms state plainly that Anthropic may not train models on Customer Content, Enterprise admins can set retention for conversations, projects and artifacts with every change tracked in audit logs, and the abuse-monitoring carve-outs are disclosed rather than buried. I hold it below the top because data is retained indefinitely unless an owner acts, and the retention controls explicitly do not reach Claude Managed Agents or cloud-run Claude Code — the agent features I would deploy first. 4 7 6 2 5
The Customer-Service Director
The commercial terms state Anthropic may not train models on customer content, the DPA is accepted through those terms, and Enterprise admins can set custom retention for conversations, projects and artifacts with every retention action tracked in audit logs, plus SOC 2 and ISO 27001 stated. Safety-flagged and user-reported conversations remain excepted, and we found no public information on zero-data-retention terms for the Work product. 2 3 4 5 6 7
The Works Council Member
The commercial terms state that Anthropic may not train models on Customer Content, a data processing agreement is accepted through those same terms, and Enterprise admins can set custom retention for conversations, projects and artifacts with every change tracked in audit logs and flagged-content retention spelled out in writing — most of what a works council needs on paper. The default is indefinite retention, however, and the retention controls exclude Claude Managed Agents and any feature running Claude Code in the cloud, exactly the agent surfaces where staff activity would accumulate. 2 4 6 7
The Skeptic
The commercial terms bar training on Customer Content and a DPA is accepted with those terms, Enterprise admins can set retention from a 30-day floor with permanent deletion and audit logging, and SOC 2 and ISO 27001 are stated. It stops short of full accounting: retention is indefinite by default, the controls exclude Claude Design, Claude Tag, Claude Managed Agents and features running Claude Code in the cloud, and the privacy policy says safety-flagged conversations keep being used for model improvement even after an opt-out. 4 6 7 5 2
Models & processing location
Show reasoningHide reasoning
How this is scored
Which models sit behind the product and who runs them, where inference and storage happen, whether the customer can pin processing to the EU, choose or exclude model providers, bring its own model or key, or run the product in its own cloud or on premises.
0 — The models behind the product and the processing location are not stated.
3 — The model family is named, but where inference runs and through which provider (the model maker, a hyperscaler, the vendor) is unstated.
5 — Models and their providers named, with an EU processing option stated for storage and at least the default model.
8 — Each available model listed with provider and inference region, EU-only processing selectable by the admin and enforced across models, model providers excludable, and bring-your-own-key or own-cloud deployment documented.
10 — Processing fully under the customer's control and documented: EU-only or on-premises inference for every feature including agents and connectors, open-weight or self-hosted models supported, data flows per feature published, and a residency commitment in the contract rather than on a marketing page.
The Head of IT
The model lineup is named — Opus, Sonnet, Haiku and Mythos, with Opus on paid plans only — and the privacy policy states that personal data is transferred to servers in the US or other countries outside the EEA and the UK. Claude is listed as available through AWS, Google Cloud and Microsoft Foundry, but the only regional pointer, to the Amazon Bedrock user guide, lists no specific regions on the captured page. We found no public information on an EU processing option, admin-selectable residency, bring-your-own-key, or an own-cloud or on-premises deployment for these plans. 1 3 5 8
The Data Protection Officer
The model families are named — Opus, Sonnet, Haiku and Mythos — but the privacy policy states data is transferred to servers in the US or to other countries outside the EEA and the UK, with no EU processing option documented for the service, and the AWS region page points to the Bedrock user guide without listing regions. Anthropic runs its own models, so there is no third-party routing to govern, but I found no public information on an EU-only inference setting, provider exclusion, bring-your-own-key, or deployment in the customer's own cloud for the Work plans. 1 3 5 8
The Automation Lead
The model families are named — Mythos including Fable, Opus, Sonnet, Haiku — with per-role control over which models members may use, and the privacy policy states personal data is transferred to servers in the US or other countries outside the EEA and the UK. We found no public information on pinning this product's inference or storage to the EU; the regions page captured simply hands off to Amazon Bedrock's own documentation. 1 3 5 8
The Customer-Service Director
The Claude model families (Opus, Sonnet, Haiku, Mythos) are named, run by the vendor itself and also reachable through AWS, Google Cloud and Microsoft Foundry, but we found no public information on where inference and storage happen for these plans or on an EU processing option an admin can pin. The privacy policy places personal data on US servers. 1 3 5 8
The Works Council Member
The model families are named and the privacy policy states plainly that personal data is transferred to servers in the US or to other countries outside the EEA and UK, so the processing location is documented but not European, and we found no public information on an EU processing option for the hosted service. For the Amazon Bedrock route, the page defers to AWS's own user guide and lists no regions itself. 1 5 8
The Skeptic
The models are named (Opus, Sonnet, Haiku, Mythos) and the vendor runs them itself, but the privacy policy states personal data is transferred to servers in the US, we found no EU processing option for this product, and the Amazon Bedrock regions page captured here lists no specific regions. A buyer wanting inference pinned to the EU finds nothing to hold on to. 1 5 8 2
Admin, identity & access controls
Show reasoningHide reasoning
How this is scored
What the IT department controls: SSO and SCIM provisioning, roles and groups, which models, agents and connectors each group may use, retrieval that respects the permissions of the source system, audit logs, usage reporting, and policies such as data-loss prevention — and whether reporting on staff use is scoped and documented.
0 — No admin features documented; accounts are individual sign-ups.
3 — A team workspace with an owner and member invitations, but no SSO, no roles and no audit log.
5 — SSO, admin and member roles, user management, and an admin view of usage.
8 — SAML/OIDC SSO with SCIM provisioning, group-based permissions for models, agents and connectors, retrieval that inherits source-system permissions, exportable audit logs, and usage analytics with documented scope.
10 — Governance a regulated company can operate: fine-grained policies (DLP, content filters, prompt and output logging rules), audit logs streamed to a SIEM, per-agent ownership and approval, usage reporting that can be aggregated or pseudonymised for works-council agreements, and every admin action logged.
The Head of IT
This covers most of what an eight-hundred-seat rollout needs: SAML single sign-on with domain capture, SCIM and just-in-time provisioning, and group-based custom roles that decide which models a role may use — with an effort-level cap and default model — which connectors it may reach at per-connector levels from Blocked to Always allow, and which admin areas from Billing to Privacy it manages, with groups synced from the identity provider via SCIM. Audit logs, usage analytics, spend controls with per-group monthly limits, and a Compliance API are listed for the Enterprise plan. We found no public information on data-loss-prevention policies, streaming audit logs to a SIEM, or pseudonymised reporting for works-council agreements, and permission-respecting retrieval is evidenced only for the Pipedrive CRM connector, not for SharePoint. 2 10 11 12
The Data Protection Officer
Enterprise gets SSO/SAML with domain capture, SCIM or JIT provisioning with identity-provider switching documented, custom roles governing capabilities, connectors and which models each role may use, per-group spend limits, audit logs and usage analytics, with retention actions tracked in the audit log. Connector tools can be set to need approval per call, the 'always allow' setting for write-capable tools is off by default, and the Pipedrive connector is documented to respect existing user permissions. I found no public information on audit-log export or streaming to a SIEM, data-loss prevention, or pseudonymised reporting for works-council agreements. 2 10 11 7 12
The Automation Lead
This is what I want before wiring an agent into company systems: SSO with SAML and domain capture, SCIM provisioning, custom roles with per-connector permission levels — every tool 'Always allow', 'Needs approval', 'Blocked' or per-tool 'Custom' — plus model access per role and per-group monthly spend limits. The gaps sit at the regulated end: we found no public information on data-loss prevention rules, streaming audit logs to a SIEM, or per-agent ownership, and the scope of the usage analytics is not described. 2 10 11 12
The Customer-Service Director
Enterprise brings SSO with domain capture, SCIM and JIT provisioning, audit logs, and an unusually fine-grained role model: per-role model access with effort caps, per-connector permission levels including per-call approval, admin areas split view/manage, and per-group spend limits. We found no public information on exporting audit logs to a SIEM, data-loss-prevention policies or per-agent ownership and approval. 2 3 10 11 12
The Works Council Member
Enterprise plans carry single sign-on with domain capture, SCIM provisioning, audit logs and usage analytics, and the role-based permissions documentation is unusually thorough: groups synced from the identity provider, per-role model access, connector permission levels that default to "Needs approval", per-group spend limits, and retrieval through the Pipedrive connector that respects existing user permissions. For a works-council agreement I would still need usage reporting that can be aggregated or pseudonymised, and we found no public information on either that or on streaming audit logs to a SIEM. 2 7 10 11 12
The Skeptic
SAML SSO with domain capture, SCIM and JIT provisioning, custom roles governing models, capabilities and connectors per group, connector calls set to need approval by default for new roles, and audit logs on Enterprise — the Pipedrive connector even respects source permissions. We found no public information on DLP policies, SIEM streaming of logs, or usage reporting that can be pseudonymised, which is what keeps a regulated buyer from the top rung. 2 10 11 12 7
Integrations, connectors & knowledge
Show reasoningHide reasoning
How this is scored
How the product reaches company knowledge and systems: connectors to file stores, wikis, mail, CRM, ticketing and ERP, how fast and how completely they sync, the API, support for open protocols such as MCP, and how the assistant cites the sources it answers from.
0 — No integrations; users paste text or upload files one at a time.
3 — File upload and one or two connectors (for example a drive), with no API and no statement on sync or citations.
5 — A connector library covering the common file stores, wikis and collaboration tools, answers citing sources, and a documented API.
8 — Connectors to file stores, wikis, mail, CRM and ticketing with incremental sync, custom connectors or MCP support, an API and webhooks for embedding assistants and agents in other tools, and source citations on every grounded answer.
10 — The product sits in the company's systems: read and write connectors with documented scopes, MCP servers and clients, contact-centre or telephony integration where relevant, a knowledge layer that can be audited and corrected, and the data, agents and configuration exportable when the customer leaves.
The Head of IT
Connectors for Gmail, Google Drive, Slack and Microsoft 365 are named and available on all plans, remote MCP is supported with a marketplace — the official Pipedrive MCP connector ships seventy-nine read-and-write CRM tools over OAuth and respects existing user permissions — and the Claude API, plugins and webhooks are documented. We found no public information on source citations for grounded answers, on sync speed or completeness, or on ticketing and ERP connectors. 1 2 3 5 12
The Data Protection Officer
Connectors cover Gmail, Google Drive, Slack and Microsoft 365, remote MCP connectors are supported through a marketplace that includes an official Pipedrive connector with read and write CRM tools over OAuth that respects existing user permissions, and the Claude API plus webhooks and external APIs are documented. I found no public information on how fast or how completely connectors sync, or on how the assistant cites the sources it answers from. 1 2 5 12
The Automation Lead
Remote MCP connectors, a documented API, webhooks and a marketplace with an official Pipedrive connector carrying seventy-nine read-and-write tools that respect existing CRM permissions — that is real plumbing, not slideware. The library as documented leans on mail, drives and collaboration tools; we found no public information on ticketing connectors, sync frequency or completeness, or how grounded answers cite their sources. 1 2 5 12 13
The Customer-Service Director
There is a genuine connector library — Gmail, Google Drive, Slack, Microsoft 365, remote MCP connectors, a documented API and webhooks, and an official Pipedrive connector with 79 read/write CRM tools that respects existing user permissions. For a contact centre we found no public information on ticketing or telephony connectors, sync speed, or how answers cite their sources. 1 2 5 12 13
The Works Council Member
Connectors reach Gmail, Google Drive, Slack and Microsoft 365, a marketplace offers MCP connectors such as a Pipedrive integration with read and write tools over OAuth, and the Claude API plus webhooks and external APIs are documented. We found no public information on sync speed or completeness, on building custom connectors, or on how the assistant cites the sources behind an answer. 1 2 5 12
The Skeptic
Connectors to Gmail, Google Drive, Slack and Microsoft 365, an official Pipedrive MCP connector with 79 read-and-write tools behind OAuth that respects existing user permissions, plus remote MCP, webhooks and a documented API. We found no public information on sync speed or completeness, and none on how answers cite their sources — the two things that decide whether this is a knowledge layer or a file drawer. 2 12 1 5
Agent capabilities & guardrails
Show reasoningHide reasoning
How this is scored
What an agent can actually do and where it stops: multi-step tasks, tool use and actions in other systems, scheduling and triggers, human approval before consequential actions, run logs, evaluation and testing before release, and the limits the vendor states rather than hides.
0 — No agents; the product is a chat window, whatever the marketing says.
3 — Saved prompts or custom assistants with instructions and files, but no actions in other systems and no run history.
5 — Agents that call tools or APIs and run multi-step tasks, built without code, with a log of each run.
8 — Agents that act in connected systems with scoped permissions, human approval configurable per action, triggers and schedules, test and evaluation tooling before release, and documented limits and failure handling.
10 — Agents a business can put in production and account for: versioning and staged rollout, evaluation against test sets, monitoring of quality and cost per agent, handover to a human with full context, every action attributable to the agent and the person who deployed it, and the vendor's own statement of what agents must not be trusted with.
The Head of IT
The managed-agents documentation is unusually candid about mechanics and limits: workflows run in the background with session event streams, fan out to up to 1,000 agents and 64 concurrent threads per run, default to 24-hour lifetimes, pause on budget, usually retry a failed agent once, and connector tool calls can sit behind per-role approval levels with Needs approval as the default for new roles. The same pages state plainly that there is no approval step before a run starts. We found no public information on evaluation or test tooling, agent versioning, staged rollout, or per-agent quality and cost monitoring. 15 11 2 5
The Data Protection Officer
Managed agents run multi-step workflows in the background with concurrent agents, documented limits (24-hour run lifetime by default, up to 1,000 agents per run), failure retry, and run events on a session event stream, while connector calls can be gated per action at 'needs approval'. Scheduled tasks appear on paid plans. The documentation is candid that there is no approval step before a run starts, and I found no public information on test or evaluation tooling before release, or on monitoring quality and cost per agent. 15 11 2 3 5
The Automation Lead
Agents run multi-step sessions that act on your behalf in connected systems, run events stream on the session so you can follow each step, and the documentation is unusually honest about limits — a thousand agents per run, sixty-four concurrent threads, a twenty-four-hour run lifetime, a failed agent usually retried once on a new thread. What stops me signing off: the docs state there is no approval step before a managed-agent run starts, though connector tools can be gated so members confirm each call, and we found no public information on test or evaluation tooling to rehearse an agent before the whole company uses it. 5 15 11 12 3
The Customer-Service Director
Managed agents run multi-step workflows in the background with fan-out, budgets, run states, retry on failure and honestly documented limits, and connector calls can require per-call confirmation with write tools needing approval by default. The documentation states there is no approval step before a run starts, and we found no public information on evaluation against test sets, versioned rollout, or handover to a human with full context. 2 3 4 11 15
The Works Council Member
Managed agents run multi-step workflows that act on connected systems under scoped permissions, and the approval machinery is what a works council wants on paper: a newly created role defaults to "Needs approval" on every connector, and letting members set "Always allow" on write-capable connector tools is off by default at the organization level. Limits, failure handling and even the fact that there is no approval step before a run starts are documented candidly, but we found no public information on testing or evaluation before an agent is released, on versioning, or on monitoring cost per agent. 2 3 11 12 15
The Skeptic
Managed agent runs fan out with documented limits (up to 1,000 agents per run, 64 threads, 24-hour default lifetime), a retry on failure, event-stream logging, budget controls and scoped connector permissions — and the vendor candidly states there is no approval step before a run starts. We found no public information on test or evaluation tooling, versioning or staged rollout, so I will not call these production-grade on the evidence. 15 11 5 3
European sovereignty
panel opinion
Show reasoningHide reasoning
How this is scored
Who the contracting entity and the parent company are, where customer data is hosted, and who the subprocessors are — model providers above all, then hosting, search, telephony and email. Independently sourced by the sovereignty pipeline; for an AI product the model provider is a subprocessor, so a European interface over a non-European model counts as such.
0 — Non-EU vendor and contracting entity, hosting unstated, subprocessors (including model providers) unnamed.
3 — A privacy policy exists, but hosting location is unstated or non-EU and the model providers behind the product are not named as subprocessors.
5 — EU hosting stated and an EU contracting entity, but prompts reach non-EU model providers or other subprocessors without a stated safeguard or EU region.
8 — EU hosting on named infrastructure, EU contracting entity, every subprocessor listed with its location including model providers, EU inference available, and a DPA offered.
10 — Sovereign end to end and evidenced: European ownership, entity, hosting and model processing, every subprocessor European or optional, data export and deletion documented, and a security or certification statement published rather than implied.
The Head of IT
European customers contract with and are controlled by Anthropic Ireland Limited, but the parent is the US company Anthropic PBC, the models behind the service are the vendor's own, and the privacy policy states personal data is transferred to servers in the US or other countries outside the EEA and the UK, with standard contractual clauses carrying the transfers. A subprocessor list is published by pointer. We found no public information on EU hosting or EU inference. 5 2 4
The Data Protection Officer
EEA, Swiss and UK customers contract with Anthropic Ireland Limited in Dublin, a subprocessor list is published, and standard contractual clauses are the stated transfer safeguard — but the privacy policy states data is transferred to servers in the US or other countries outside the EEA and the UK, and Anthropic's own US models are the inference layer with no EU region documented. I found no public information on EU hosting for the service; arbitration for European customers sitting in Dublin is a small comfort. 5 4 2
The Automation Lead
EEA, Swiss and UK customers contract with Anthropic Ireland Limited, with Dublin arbitration and standard contractual clauses covering transfers, but the privacy policy states personal data is processed on US servers, so hosting is not European. A subprocessor list is published as a reference, though the pages captured name only two Korean processors, and we found no public information on EU inference or an EU hosting option for this product. 5 2 4 6
The Customer-Service Director
EEA, Swiss and UK customers contract with Anthropic Ireland Limited, a subprocessor list is published, and transfers rely on standard contractual clauses. The privacy policy places personal data on US servers, the models behind the product are run by a US parent, and we found no public information on EU hosting or inference. 2 4 5
The Works Council Member
EEA, Swiss and UK customers contract with Anthropic Ireland Limited, a subprocessor list is published, and standard contractual clauses are the stated transfer safeguard — but the model maker and parent is Anthropic PBC in the US, and the privacy policy says personal data goes to US servers or other countries outside the EEA and UK. We found no public information on EU hosting or EU inference for the product itself. 2 4 5
The Skeptic
EEA, Swiss and UK customers contract with Anthropic Ireland in Dublin and arbitrate there, but the parent is the US public benefit corporation, personal data is stated to be transferred to US servers or countries outside the EEA, and the model provider is that same US company with no EU inference documented. SCCs and a pointer to a published subprocessor list are the whole safeguard story; that is a compliance wrapper, not sovereignty. 5 4 2
Pricing transparency
Show reasoningHide reasoning
How this is scored
Whether a company can compute the real annual cost from public pages alone: price per seat, per agent, per conversation or per usage unit, what is included in each tier, usage limits and overage, which models, connectors or agents cost extra, minimum seats and term.
0 — No public prices at all; every customer is a sales call.
3 — A starting price or "from" figure exists, but usage limits, included models and what scales the cost are unclear.
5 — Prices per tier are public with the main inclusions, but usage limits, overage, agent runs or enterprise features (SSO, audit logs, EU processing) are unpriced or "on request".
8 — Every tier priced publicly with seat or usage units, usage limits and overage rates, add-ons listed, minimum seats and term stated.
10 — Complete price computability: annual cost derivable for a given number of seats, agents or conversations, with every usage rate, model surcharge, add-on and support tier and the cancellation and data-return terms published.
The Head of IT
Enterprise is published at $20 per seat per month plus usage billed at API rates, with a 20-seat minimum and annual commitment, and the tiers below are public with their inclusions — Pro at $17 per month with annual subscription discount ($200 billed up front) or $20 if billed monthly, Max from $100 per month. But usage limits are described only relatively, the usage rates sit on a separate Model Pricing Page the captured pages reference rather than show, and we found no public information on a Team seat price, so the real annual cost for eight hundred seats cannot be computed from the captured pages. 3 2 4
The Data Protection Officer
Free at $0, Pro at $17 per month with the annual subscription discount ($200 billed up front) or $20 if billed monthly, Max from $100 per month, and Enterprise at $20 per seat per month plus usage billed at API rates with a 20-seat minimum and annual commitment are published, with taxes excluded and rate changes on 30 days' notice. Usage is priced by reference to a separate Model Pricing Page, and I found no public information quantifying usage limits, overage rates, or a price for the Team plan. 3 2 4
The Automation Lead
The headline numbers are public: the free plan at $0, Pro at $17 per month with annual subscription discount ($200 billed up front) or $20 if billed monthly, Max 'From $100 Per month', and Enterprise at $20 per seat per month plus usage billed at API rates with a minimum of 20 seats and an annual commitment. A buyer still cannot total a year from these pages alone: usage limits are flagged but not quantified, the per-token rates live on a separate pricing page, and we found no public information on a Team plan price, overage rates or add-ons. 3 2 4 15
The Customer-Service Director
Five tiers are priced publicly with per-tier feature matrices: Free at $0, Pro at "$17 Per month with annual subscription discount ($200 billed up front)" or $20 monthly, Max from $100 per month, and Enterprise at $20 per seat per month billed annually with a 20-seat minimum and annual commitment. Usage beyond that is billed at API rates with only "Usage limits apply" stated, and we found no per-conversation price or overage rates, so an annual contact-centre cost is not computable from these pages. 1 2 3 4
The Works Council Member
Seat prices are public — $0 Free, "$17 Per month with annual subscription discount ($200 billed up front). $20 if billed monthly." for Pro, Max "From $100 Per month", and "$20 per seat per month plus usage billed at API rates" for Enterprise with a 20-seat minimum and annual commitment — and agent runs are clarified as carrying no price of their own, billed at each model's rates. We found no public information on the Team seat price, on the usage limit figures, or on the API rates themselves, so a buyer cannot compute a full annual cost from these pages alone. 2 3 4 15
The Skeptic
Seat prices are public — Free at $0, Pro at $17 per month on annual subscription or $20 monthly, Max from $100 per month, Enterprise at $20 per seat per month plus usage billed at API rates with a 20-seat minimum and annual commitment. But usage limits are only relative ("Choose 5x or 20x more usage than Pro"), the API rates live on a separate Model Pricing Page, and we found no overage rates — so a year's real cost is not computable from the pages captured here. 3 2 4 1
European sovereignty — proven facts
0 of 4 dimensions provenBuilt only from facts shown on the vendor's own pages. A dimension we could not prove is left open, not scored as zero.
| Legal entity | Not determined ⚠ unverified | — | uncited Report an error |
|---|---|---|---|
| Ownership | Not determined | — | uncited Report an error |
| Data residency | Not determined ⚠ unverified | — | uncited Report an error |
| Subprocessors | Not determined | — | uncited Report an error |
Questions about Claude for Work
Is Claude for Work European?
No. whats-best.ai classifies Claude for Work as Rest of world.
Where does Claude for Work store data?
Not stated on the vendor’s pages as of 26 Oct 2026.
Does Claude for Work have a free plan?
Yes. Claude for Work has a free plan, per the vendor’s page captured 26 Oct 2026.
What are the best alternatives to Claude for Work?
As of 28 Oct 2026, the panel rates Langdock (4/5), Dust (3.5/5) and Unique AI (3.5/5) highest among the other 15 AI Agents & Assistants products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European alternative is Langdock (EU-Made). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: Claude for Work alternatives.
Where this could be wrong
- Evidence ages. The oldest capture behind this page is from 26 Oct 2026. Vendors change pricing and policies without notice; every fact reflects its source as of the capture date shown in the registry.
- Weak sourcing — Legal entity, Data residency. Not confirmed on the vendor’s own pages as captured.
- AI can misread a source. Extraction and judgement are automated; a citation guarantees traceability, not infallibility. If something here is wrong, say so — no account needed, every report is decided within 5 business days, and accepted corrections are published.
What we left out
A claim that does not survive our checks costs us the claim, not the page. This is what was taken off this one.
- 13 product facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 7 compliance facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 5 legal facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 1 integrations fact could not be confirmed on the vendor’s page as captured and was left out of this page and of the panel’s material. Know more? Tell us
- 1 pricing fact could not be confirmed on the vendor’s page as captured and was left out of this page and of the panel’s material. Know more? Tell us
- 1 subprocessors fact could not be confirmed on the vendor’s page as captured and was left out of this page and of the panel’s material. Know more? Tell us
- 2 sovereignty dimensions could not be confirmed on the vendor’s own pages and are shown as unknown. Know more? Tell us
Sources (15)
The pages every claim on this page was read from — each one checked, dated, and kept verifiable.
- 1 Vendor homepage claude.com Checked 26 Oct 2026 Details →
- 2 Features page (Enterprise plan) claude.com Checked 26 Oct 2026 Details →
- 3 Pricing page claude.com Checked 26 Oct 2026 Details →
- 4 Terms of service www.anthropic.com Checked 26 Oct 2026 Details →
- 5 Privacy policy www.anthropic.com Checked 26 Oct 2026 Details →
- 6 Data use, training & retention — found from sitemap support.claude.com Checked 26 Oct 2026 Details →
- 7 Data use, training & retention — found from sitemap support.claude.com Checked 26 Oct 2026 Details →
- 8 Models & processing location — found from sitemap support.claude.com Checked 26 Oct 2026 Details →
- 9 Models & processing location — found from sitemap platform.claude.com Checked 26 Oct 2026 Details →
- 10 Admin, identity & access controls — found from sitemap support.claude.com Checked 26 Oct 2026 Details →
- 11 Admin, identity & access controls — found from sitemap support.claude.com Checked 26 Oct 2026 Details →
- 12 Integrations, connectors & knowledge — found from sitemap claude.com Checked 26 Oct 2026 Details →
- 13 Integrations, connectors & knowledge — found from sitemap claude.com Checked 26 Oct 2026 Details →
- 14 Agent capabilities & guardrails — found from sitemap platform.claude.com Checked 26 Oct 2026 Details →
- 15 Agent capabilities & guardrails — found from sitemap platform.claude.com Checked 26 Oct 2026 Details →