AI Agents & Assistants
Unique AI
UK / wider Europe Report an errorPanel rating · 6 judges · How to read the stars
Category median
Sovereignty: 1 of 4 dimensions proven
0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.
by Unique AG · www.unique.ai
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.
Unique AI at a glance
- Vendor
- Unique AG
- 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
- Not stated on the vendor’s pages as of 26 Oct 2026
- Panel verdict
- The AI panel rates Unique AI 3.5/5, the mean of 6 judges. Unique AI is a Swiss agentic platform for wealth and asset managers. 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
Unique AI is a Swiss agentic platform for wealth and asset managers. Its strongest area is data use and training, scoring 7-9: no data is used for AI model training, retention is documented store by store with deletion on request, Azure OpenAI processing is in memory only under a Microsoft opt-out with abuse monitoring disabled, and a DPA with SCCs backs ISO 27001, ISO 9001, ISO 42001 and SOC 2 Type 2 certifications. Admin and access scores run 6-8, credited to custom roles with transitive inheritance, implicit audit logging and an MCP hub scoping each agent's reach; we found no public information on single sign-on or SCIM provisioning. Sovereignty scores 5-6, the lowest of any counted criterion: the model layer rides on Microsoft, a US-headquartered provider subject to the US CLOUD Act, subprocessors are published only as categories, and the captured pages give different figures for retention and different statements on data location. Processing and models, integrations and knowledge, and agent capabilities hold at 6-7; scores never spread more than two points on any criterion, and no prices are published.
Speaks for it
- No data is used for AI model training, with the Microsoft path covered by an opt-out agreement, disabled abuse monitoring and in-memory-only Azure OpenAI processing.
- Retention is documented per data store with deletion on request and hard deletion from the database.
- Certified ISO 27001, ISO 9001 and ISO 42001, with SOC 2 Type 2 (February 2025), stated adherence to the FINMA 2018/3 Outsourcing Circular, and a published DPA with SCCs.
- Custom roles with transitive inheritance, per-resource and per-action allow and deny policies, and audit logging written implicitly on every operation.
- Three deployment models including a customer-managed tenant on the customer's own cloud, with self- or cloud-hosted custom models connectable.
Held against it
- Subprocessors are published only as categories, and we found no public information naming individual subprocessors or their locations.
- The model path runs through Microsoft, a US-headquartered provider subject to the US CLOUD Act regardless of server location, with OpenAI access through data centres in Switzerland, Europe, the US, UK and Asia.
- We found no public information on single sign-on or SCIM user provisioning.
- We found no public information on connectors for mail, CRM, ticketing or ERP systems, on webhooks, or on how grounded answers cite their sources.
- We found no public information on human approval before consequential agent actions, or on triggers and schedules for agent runs.
Best for
- You are a wealth or asset manager who needs written no-training commitments and store-by-store retention documentation for compliance review.
- Your team must run the platform in its own cloud and connect self- or cloud-hosted models.
- You need role-level control over which systems, tools and datasets each agent may touch.
- Your knowledge lives in SharePoint or Confluence and you want incremental, configurable synchronization.
Avoid if
- You need no US-headquartered provider in the processing chain, since the default model path runs through Microsoft.
- Your compliance regime requires pinning inference to Europe, as the documented OpenAI path spans Microsoft data centres in Switzerland, Europe, the US, UK and Asia.
- You operate outside the EEA and Switzerland and want a European controller, since the documented controller there is Unique Applications Inc. of New York.
- You must review a named subprocessor list with locations before signing, as the published information names only categories.
The scores
| Criterion | Panel median |
|---|---|
| Data use, training & retention | 4.5 of 5 |
| Models & processing location | 3.5 of 5 |
| Admin, identity & access controls | 3.5 of 5 |
| Integrations, connectors & knowledge | 3.5 of 5 |
| Agent capabilities & guardrails | 3.5 of 5 |
| European sovereignty | 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
For a regulated rollout the data story is unusually complete: no data is used for model training, the Microsoft path runs under an opt-out agreement with abuse monitoring disabled and in-memory-only processing, and retention is documented per data store with deletion on request, hard deletion from the database, and a DPA with SCCs behind SOC 2 Type 2 and ISO 27001 and 42001 audits. It stops short of the top level because retention is not admin-configurable — deletion is a manual request confirmed by email — and the per-provider accounting covers the Azure OpenAI path, with no equivalent statement for customer-connected models. 3 4 6 7
The Data Protection Officer
Training exclusion is stated for the vendor and, through a documented Microsoft opt-out agreement with abuse monitoring disabled, for the model provider itself; retention is published per service — prompts and answers held with deletion on request, thirty days after termination by default, ten years for audit logs — and a data processing addendum plus SOC 2 Type 2 and ISO 27001/9001/42001 audits cover the product. The captured pages give different figures for default retention (ten years after the last transaction on the privacy page versus thirty days after termination in the security documentation), and deletion runs through an email request confirmed by email rather than an admin-controlled audit trail, which keeps it just short of the top level. 3 5 6 7
The Automation Lead
No-training is stated for the vendor and for Microsoft behind the Azure OpenAI models, with an opt-out agreement, disabled abuse monitoring and in-memory-only processing described; retention is published per service for prompts, answers, documents and embedding vectors with deletion on request and hard deletion from the database, the DPA leaves all client data the client's property, and ISO 27001, ISO 42001 and SOC 2 Type 2 audits cover the product. It stops just short of the top because retention is not admin-configurable — deletion goes through an emailed request to the security team, and multi-tenant audit logs are capped at 90 days unless the contract is upsold. 3 4 6 7 11
The Customer-Service Director
The strongest data-use documentation I have seen from a vendor: no data is used for model training, Azure OpenAI processes documents in memory only with Microsoft logging and abuse monitoring disabled, and retention is spelled out per data store with deletion on request and hard deletion from the database, wrapped in a DPA and ISO 27001, ISO 42001 and SOC 2 Type 2 audits. The only gap for my compliance team is that deletion is confirmed by email to the Security and Compliance Team rather than through a self-service audit trail. 3 4 6 7
The Works Council Member
Training on staff input is ruled out in writing — none of the data is used for model training — and the named model provider is covered too: an opt-out agreement with Microsoft, abuse monitoring disabled, and in-memory-only processing with no storage. Retention is documented store by store with deletion on request and true hard deletion, though deletion runs as an emailed request rather than an admin control, and the terms reserve use of content for further development of the service while the security pages state no training occurs — the captured pages give different statements here, so we would pin this in a works agreement before rollout. 3 4 5 6 7
The Skeptic
Training and retention are documented store by store: prompts and answers in Postgres with deletion on request, documents in storage accounts, embedding vectors in Qdrant, and a flat statement that none of it is used for AI model training. The model provider side is pinned down too — Azure OpenAI processes documents in memory only, with an opt-out agreement with Microsoft and abuse monitoring disabled — and a DPA with SCCs is published, backed by ISO 27001, ISO 42001 and SOC 2 Type 2 audits. What keeps it off the top rung: deletion runs through an email to the security team rather than admin-configurable retention with a customer-facing audit trail, and the privacy policy, the retention documentation and the DPA give different figures for retention, which a buyer must reconcile. 3 5 6 7
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 default models are Azure OpenAI running on Microsoft Azure hosted in Switzerland and deployed separately within the Azure customer environment, and the deployment ladder — multi-tenant, single-tenant, customer-managed tenant on the customer's own cloud, plus documented self- or cloud-hosted custom models — is exactly the option I need instead of "trust us". I hold it below the higher level because the captured pages do not list each available model with its provider and inference region, show no admin-selectable EU-only enforcement across models, and give different statements on data location, from Azure Switzerland to "in Switzerland and other locations, as reasonably necessary". 3 5 6 7 8 9
The Data Protection Officer
The default models are Azure OpenAI GPT deployments kept separate from public OpenAI services, storage remains in Switzerland, three deployment models including a customer-managed tenant on the customer's own cloud are documented, and self- or cloud-hosted custom models can be connected. But the trust pages describe OpenAI access through Microsoft data centres in Switzerland, Europe, the US, UK and Asia, and we found no public information on an administrator-selectable EU-only processing switch enforced across models, agents and connectors, nor a published inference region per model. 3 6 7 9
The Automation Lead
The default path is Azure OpenAI with data stored on Microsoft Azure in Switzerland and a contractual commitment that storage remains in Switzerland, and it goes further than most: a customer-managed tenant on the customer's own cloud and self- or cloud-hosted custom language models are documented, with available models configurable per deployment. But the Microsoft OpenAI path is described as global — data centres in Switzerland, Europe, the US, UK and Asia — and we found no public information on an admin-selectable Europe-only inference option or a published list of every model with its inference region. 3 6 7 8 9
The Customer-Service Director
Azure OpenAI through Microsoft is the named default route, with the models deployed inside the Azure customer environment and data centres named including Switzerland, plus storage held to Switzerland for hosted deployments. Deployment is genuinely flexible — multi-tenant, single tenant, or customer-managed in our own cloud, and self- or cloud-hosted custom models can be connected — but the privacy policy also allows data in 'other locations, as reasonably necessary', and I found no admin-enforced EU-only setting or a per-model list with inference regions. 3 5 6 7 9
The Works Council Member
The default models are Azure OpenAI on Microsoft infrastructure with storage stated to remain in Switzerland, and a customer-managed tenant on the customer's own cloud plus self- or cloud-hosted custom models are documented. What we could not find is an admin setting pinning inference to Europe — the captured pages say OpenAI access runs through data centres in Switzerland, Europe, the US, UK and Asia — and no per-model inference-region list is published. 3 6 7 8 9
The Skeptic
The default brain is named: OpenAI reached through Microsoft infrastructure, storage contractually kept in Switzerland, and Azure OpenAI models deployed separately within the customer's Azure environment rather than through public OpenAI services. Three deployment models are documented, including a customer-managed tenant on the customer's own cloud, and self- or cloud-hosted custom models can be connected for standard use cases, with model availability configurable per deployment. We found no public information on a per-model list with inference regions, on an admin-selectable EU-only setting enforced across all models, or on excluding a specific provider. 3 6 7 9
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
Role-based governance is genuinely deep: custom roles with transitive inheritance, allow and deny policies over resources including assistants, scopes and MCP connectors, MCP Hub control over which systems each agent may reach, implicit audit logging of every operation with documented retention and export paths, SharePoint permission synchronisation, and per-user token and spend tracking with daily limits. Two findings hold it back for an eight-hundred-seat deployment: we found no public information on single sign-on or SCIM provisioning, and permission-aware retrieval is documented for SharePoint while the Confluence connector is documented as content-sync only. 2 8 10 11 12 13
The Data Protection Officer
Granular custom roles with transitive inheritance, per-resource and per-action allow/deny policies covering chat, assistants, MCP connectors and analytics, audit logs written implicitly on every operation and exportable on request, SharePoint permissions synchronised into the knowledge base, and usage analytics with per-user daily spending limits make this genuine IT-grade governance. We found no public information on single sign-on or SCIM user provisioning, and granular policies are manageable only via the API while the interface offers inheritance only. 8 10 11 13
The Automation Lead
The Gatekeeper permission model is genuinely fine-grained: custom roles with transitive inheritance, allow/deny policies per resource and action covering chats, assistants, scopes, MCP connectors and analytics, plus an MCP Hub where admins control which systems, tools and datasets each agent may touch. Audit logs are written implicitly on every operation and exportable on request, admins get per-user and per-assistant token spend with daily spend limits, and SharePoint retrieval can synchronize source permissions. We found no public information on SSO with SCIM provisioning, SIEM streaming of audit logs, or DLP and content-filter policies — the audit documentation states content filtering is not its goal. 2 8 10 11 13
The Customer-Service Director
My IT team gets custom roles with inheritance, API-level policies down to read, write and delete per resource, SharePoint permissions synced into the knowledge base, implicit audit logs with export paths, and admin dashboards tracking token spend per user and assistant with daily spending limits. I found no public information on single sign-on or SCIM provisioning, and policies created via the API do not show their detail in the UI, which my admins would grumble about. 2 8 10 11 13
The Works Council Member
The permission model is unusually fine-grained — custom roles with inheritance, per-resource policies over chat, analytics, agents and connectors, and an MCP Hub where admins control what each agent may touch — and audit logs with user and machine attribution are documented down to their limitations. We found no public information on single sign-on or automated user provisioning, the Confluence connector syncs content without source-system permissions, and nothing documents usage reporting that can be aggregated or pseudonymised for a works-council agreement, which is the piece we need most. 2 8 10 11 12 13
The Skeptic
Access control is genuinely engineered: custom roles with transitive inheritance, resource-by-action permissions (read, write, delete and so on) settable through the Gatekeeper API, implicit audit logging on every operation, SharePoint permissions synced into the knowledge base, and per-user token and spend tracking with daily limits. The unusually honest documentation of limits — audit event lists non-deterministic, exports prepared over days and capped at five business days per request, granular policies invisible in the UI — is to the vendor's credit. We found no public information on single sign-on or SCIM provisioning, which for a bank buying a thousand seats is the first question IT asks. 2 8 10 11 13
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
SharePoint and Confluence connectors are documented to an operational depth I can hand my team — incremental file-diff sync, fifteen-minute scheduled scans, MIME filtering, per-site access grants — and MCP servers, the MCP Hub and APIs with an SDK for embedding search and Q&A into internal tools are documented. We found no public information on mail, CRM or ticketing connectors, on webhooks, or on how grounded answers cite their sources, which keeps the connector story short of a broad library. 2 9 12 13
The Data Protection Officer
Deep-documented connectors for SharePoint and Confluence with incremental file-diff sync, scans defaulting to every fifteen minutes, optional SharePoint permission synchronisation, configurable file types and rate limits, plus an MCP hub with custom MCP servers and a documented API and SDK for embedding assistants in internal systems. We found no public information on connectors for mail, CRM, ticketing or ERP, on webhooks, or on how grounded answers cite their sources. 2 12 13
The Automation Lead
The SharePoint and Confluence connectors are documented in real depth — incremental file-diff sync, permission synchronization, selective flag-based ingestion, configurable 15-minute scans — and MCP is a first-class citizen with custom MCP servers, an MCP Hub and MCP-scoped permissions, alongside GraphQL APIs and an SDK for embedding search and workflow automation. We found no public information on connectors to CRM, ticketing or ERP systems, on webhooks, or on how answers cite their sources. 1 2 9 12 13
The Customer-Service Director
The SharePoint and Confluence connectors are properly engineered — incremental file-diff sync, configurable 15-minute scans, permission sync from SharePoint — and the MCP hub with custom MCP servers plus APIs and an SDK let us embed agents in our own tools. For a contact centre, though, I found no public information on CRM or ticketing connectors, webhooks, telephony integration, or how answers cite their sources. 1 2 12 13
The Works Council Member
SharePoint and Confluence connectors are documented in depth with incremental sync, SharePoint permissions carry over into retrieval, and MCP servers, a governed integration layer and an API with SDK round out the picture. We found no public information on mail, CRM or ticketing connectors, on webhooks, or on citations showing the source behind each grounded answer. 1 2 7 12 13
The Skeptic
The SharePoint and Confluence connectors are documented in real depth — incremental file-diff sync, permission synchronization for SharePoint with granular Microsoft Graph scopes, label-driven ingestion, rate limits and configurable scans every 15 minutes by default — and custom MCP servers plus a GraphQL API and SDK open the platform to internal systems. The documented library reaches file stores and wikis; we found no public information on mail, CRM or ticketing connectors, on webhooks, or on how a grounded answer cites its sources — the question I always ask before trusting a RAG answer. 2 9 12 13
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
Prebuilt agents for KYC, RFP analysis and due diligence plus no-code custom agents in Spaces are documented, with agent reach scoped through the MCP Hub, a four-stage evaluation funnel, benchmarking against human-verified ground truths with quality-risk flags and human review, and structured per-tool execution-time logging for run visibility — evaluation tooling most vendors only promise. We found no public information on human approval before consequential actions or on triggers and schedules for agent runs. 2 8 14 15
The Data Protection Officer
Prebuilt finance agents for KYC, RFP and onboarding alongside no-code custom agents in Spaces, multi-step workflows with per-tool execution timing and an in-loop hallucination check, agent access to systems and datasets governed per agent through the MCP hub, and benchmarking against human-verified ground truths where flagged outputs go to a human reviewer. We found no public information on human approval configurable before consequential actions, or on schedules and triggers for agent runs. 1 2 14 15
The Automation Lead
Agents run multi-step loops with per-tool execution timing logged, act only within admin-scoped permissions, and there is real pre-release testing: automatic and manual benchmarking against human-verified ground truths, quality-risk flagging routed to a human reviewer, and a hallucination check inside the loop. We found no public information on configurable human approval before an agent takes a consequential action, or on triggers and schedules for agents — the two guardrails I would want before letting one loose on an ERP. 1 2 8 14 15
The Customer-Service Director
Agents run multi-step loops with per-tool execution timing in their run output, tool access is governed per agent through the MCP hub, and pre-release evaluation is genuinely strong — benchmarking against human-verified ground truths with quality-risk flags and a human reviewer making final calls. What I could not find is anything on human approval before consequential actions, triggers or schedules, or handover to a human with the full conversation context, all of which I would need proven before this touches a live caller. 2 14 15
The Works Council Member
Agents are built without code, act through tools under admin-set permissions, and the evaluation story is real: benchmarking against human-verified ground truths with a human reviewer on flagged outputs, plus per-step execution logging for every run. What we could not find is a documented human-approval gate before an agent takes a consequential action, or triggers and schedules — for us that gate must be in writing before any rollout. 2 8 11 14 15
The Skeptic
These are real agents, not saved prompts: an agentic loop with planning, tool execution and post-processing, prebuilt finance agents for KYC, RFP and due diligence workflows, and benchmarking against human-verified ground truths with quality-risk flagging and a four-stage model evaluation funnel, plus per-step execution-time logging. The MCP Hub lets an admin scope which systems, tools and datasets each agent may touch. We found no public information on human approval before consequential actions, on agent triggers or schedules, or on versioning and staged rollout — so I cannot say what stops an agent from acting unsupervised. 1 2 8 14 15
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
EEA and Swiss customers contract with Unique AI AG of Zurich under Swiss law, hosting sits on Microsoft Azure in Switzerland with the storage committed to remain in Switzerland in the DPA, and the Microsoft model path carries stated safeguards: a logged-out deployment with no storage, an abuse-monitoring opt-out and SCCs. It stays below the strong level because subprocessors are published only as categories rather than a named list with locations, the privacy policy permits data "in Switzerland and other locations, as reasonably necessary", and customers outside the EEA and Switzerland have a New York entity as controller. 3 4 5 6 7
The Data Protection Officer
The contracting controller for EEA and Swiss customers is Unique AI AG of Zurich, storage remains in Switzerland on Microsoft Azure under Swiss law and Zurich jurisdiction, and the model provider — Azure OpenAI, deployed separately within the customer's Azure environment — is named with its region. However, we found no public information naming the individual subprocessors or their locations (only categories such as IT services, telecommunications and consulting), Microsoft remains a US-headquartered provider subject to the US CLOUD Act regardless of server region, a New York entity is controller for customers outside the EEA, and ownership is not stated. 3 5 6 7
The Automation Lead
The contracting entity for Europe is Unique AI AG in Zurich, hosting is on named infrastructure — Microsoft Azure in Switzerland — with a contractual commitment that storage remains in Switzerland, and a DPA with standard contractual clauses is published. However, the public subprocessor information names only broad categories with no names or locations, the models sit with US-headquartered Microsoft whose OpenAI path spans data centres in the US, UK and Asia, and the captured pages give different statements on where data is maintained — the DPA says storage remains in Switzerland while the privacy policy says Switzerland and other locations as reasonably necessary. 3 4 5 6 7
The Customer-Service Director
The contracting entity is Unique AI AG in Zurich under Swiss law and Zurich courts, storage stays on Microsoft Azure in Switzerland, and the model route through Microsoft carries stated safeguards — an opt-out agreement with abuse monitoring disabled, no data shared with OpenAI's public service, and Azure data centres including Switzerland. What holds me back is that only subprocessor categories are published rather than a full list with locations, and Microsoft remains a US-headquartered provider subject to the CLOUD Act wherever its servers sit. 3 4 5 6 7
The Works Council Member
The contracting entity for EEA and Swiss customers is Unique AI AG in Zurich, storage is stated to remain in Switzerland on Microsoft Azure, and a DPA with standard contractual clauses is published. But the model provider is Microsoft, a US-headquartered company subject to the US CLOUD Act regardless of server location, the privacy policy names subprocessor categories rather than individual subprocessors with locations, and customers outside the EEA get a New York entity as controller. 3 4 5 7 8
The Skeptic
The contracting entity for EEA and Swiss customers is Unique AI AG at Stockerstrasse 34, Zurich, under Swiss law with exclusive venue in Zurich, storage remains in Switzerland on named Azure infrastructure, and a DPA with SCCs is published. The model layer rides on Microsoft, a US-headquartered provider subject to the US CLOUD Act regardless of server location, mitigated by an opt-out agreement, disabled abuse monitoring and data centres in Switzerland — and a New York entity, Unique Applications Inc., is controller for customers outside the EEA and Switzerland. Subprocessors are published only as categories; we found no public information naming individual subprocessors with their locations, including the DPA appendix. 3 4 5 6 7
Pricing transparency
not rated — the vendor publishes no price
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
Everything published about money is legal plumbing — Order Form-based subscriptions, 30-day invoicing, taxes extra, fees non-cancelable and non-refundable — and we found no public price figures, tiers, seat or usage rates. An annual cost for eight hundred staff cannot be computed from what is published; it is an Order Form negotiation. 4
The Data Protection Officer
The published terms describe billing mechanics — invoicing due in thirty days, non-refundable and non-cancelable subscriptions, automatic renewal with thirty days' notice, taxes extra — but every price sits in an order form; we found no public price per seat, per agent or per usage unit and no published tiers, and longer audit-log retention is described only as a contract upsell to a single or customer-managed tenant. A company cannot compute any annual cost from the captured pages. 4 11
The Automation Lead
We found no price figures on any captured page — every subscription is defined by an Order Form, so a company cannot compute a real annual cost from public information. The only public commercial terms are mechanics: advance billing on credit card by default, taxes extra, fees non-cancelable and non-refundable, auto-renewal with 30 days' notice, and audit-log retention beyond 90 days requiring a contract upsell to a single or customer-managed tenant. 4 11
The Customer-Service Director
There is nothing here my CFO could forecast from: subscriptions are negotiated per Order Form, fees are due 30 days from the invoice date, non-cancelable and non-refundable, and no public price per seat, per agent or per conversation appears on any page we read. Every cost question becomes a sales conversation. 4
The Works Council Member
We found no public prices at all — no seat price, no tiers, no usage limits — with access subject to a subscription under an Order Form. The commercial small print that is public is not buyer-friendly: fees are non-cancelable and non-refundable, quantities cannot decrease during the term, and subscriptions auto-renew unless written notice is given at least 30 days before the term ends. 4
The Skeptic
We found no public prices: subscription fees are set in the Order Form, and none of the captured pages — homepage, platform, trust center, terms of service, documentation — shows a per-seat, per-agent or usage price, a tier breakdown or usage limits. The published material covers billing mechanics only: invoiced fees due 30 days from the invoice date, subscription obligations non-cancelable and payments non-refundable, automatic renewal unless notice is given at least 30 days before term end, fees changeable at renewal on 30 days' written notice, taxes extra. The annual cost for a given number of seats cannot be computed from public pages. 1 2 3 4
European sovereignty — proven facts
1 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 | US CLOUD Act reach ⚠ unverified | 0/2 pts | 3 Report an error |
Questions about Unique AI
Is Unique AI European?
Yes, though not from the EU. whats-best.ai classifies Unique AI as UK / wider Europe.
Where does Unique AI store data?
Not stated on the vendor’s pages as of 26 Oct 2026.
Does Unique AI have a free plan?
Not stated on the vendor’s pages as of 26 Oct 2026.
What are the best alternatives to Unique AI?
As of 28 Oct 2026, the panel rates Langdock (4/5), Dust (3.5/5) and Omnifact (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: Unique AI 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.
- Weak sourcing — Subprocessors. Servers sit in Switzerland and Azure OpenAI deployments are logged-out, but Microsoft is a US-headquartered provider subject to the US CLOUD Act regardless of server region, and the full subprocessor list (DPA Appendix B) was not among the excerpts.
- 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.
- 9 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
- 9 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
- 2 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
- 2 support 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 data 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 www.unique.ai Checked 26 Oct 2026 Details →
- 2 Features page (platform) www.unique.ai Checked 26 Oct 2026 Details →
- 3 Trust center www.unique.ai Checked 26 Oct 2026 Details →
- 4 Terms of service www.unique.ai Checked 26 Oct 2026 Details →
- 5 Privacy policy www.unique.ai Checked 26 Oct 2026 Details →
- 6 Data use, training & retention — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 7 Data use, training & retention — found from sitemap www.unique.ai Checked 26 Oct 2026 Details →
- 8 Models & processing location — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 9 Models & processing location — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 10 Admin, identity & access controls — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 11 Admin, identity & access controls — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 12 Integrations, connectors & knowledge — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 13 Integrations, connectors & knowledge — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 14 Agent capabilities & guardrails — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →
- 15 Agent capabilities & guardrails — found from sitemap docs.unique.ai Checked 26 Oct 2026 Details →