Orchestrate every channel
from one intelligent
layer.
In 30 minutes we'll show you how Twinny connects to your operation and starts delivering results from week 1.
Twinny is built against European compliance frameworks from day one. Every conversation, every decision and every data point is auditable and reversible.
PILLARS
Not a list of promises. Each pillar has a mechanism and an audit behind it.
Default deployment on AWS Frankfurt or Ireland. On-premise option for regulated sectors. Customer data never leaves the EU without explicit consent.
We don't train generic models on your conversations, cases or documents. Models are yours, logs are yours, decisions are signed in your name.
Every agent action is logged with timestamp, applied policy, consulted data and cryptographic signature. Exportable to your SIEM in OpenTelemetry.
Roles defined by your organization. Informed consent in every conversation, GDPR rights in one click, guaranteed right to be forgotten.
TLS 1.3 in transit, AES-256 at rest. Keys rotated per policy, custody with KMS independent of the cloud provider.
Quarterly penetration tests, independent annual SOC 2 review, active bug bounty programme. Reports available under NDA.
IDENTITY, ACCESS & AUDIT
Access to the platform follows the standards your IT team already uses, without requiring exceptions to corporate policy.
SAML 2.0, OpenID Connect, Google Workspace and Microsoft Entra ID. Automatic provisioning and de-provisioning via SCIM 2.0.
Second factor by default: TOTP, WebAuthn / passkeys, push notifications. Policy configurable per role.
Roles defined by your organization, not by Twinny. Permissions at module, pipeline, agent and sensitive-data level.
Every login, every access to personal data, every configuration change. Exportable to your SIEM in a standard format.
Configurable timeouts, remote session kill, optional IP allowlisting, private VPN and AWS peering support.
Sensitive actions (mass send, agent policy change, recording access) require configurable dual approval.
CERTIFICATIONS AND FRAMEWORKS
EU AI ACT
The EU regulation classifies AI uses by risk. Twinny manages each deployment according to the level that applies and documents the required controls.
| Level | What it requires | In Twinny |
|---|---|---|
| Minimal | Unrestricted under the EU AI Act (informational chatbots). | Default |
| Limited | Transparency: the agent declares it is AI at the start of each conversation. | Active |
| High | For uses classified as high-risk (HR, credit, clinical care): technical documentation, human supervision and EU registry. | Case by case |
| Prohibited | Twinny does not operate in prohibited uses: social scoring, manipulation, real-time biometric surveillance. | N/A |
PRIVACY AND GDPR
Each commitment maps to a concrete mechanism: a signed document, a measurable process or a contact path.
Twinny operates as data processor for personal data entrusted by the customer, who remains the controller.
Data Processing Agreement available for all customers handling personal data. Templates aligned with EU Commission Standard Contractual Clauses.
Public, versioned list of sub-processors. Any change is communicated 30 days in advance.
Access, rectification, deletion, portability, opposition and limitation handled within 30 days of the request.
Customer notification within 72 hours of detection, with technical detail and remediation plan.
Data Protection Officer reachable at clientes@twinny.es for queries, requests and authorities.
FREQUENTLY ASKED
No. Customer conversations, cases and documents are not used to train generic models. Embeddings and memory stay inside the customer's tenant.
EU cloud by default (AWS Frankfurt or Ireland). On-premise option for regulated sectors. Data does not leave the EU without explicit, logged consent.
Yes. Twinny acts as data processor (GDPR art. 28) and provides a signable DPA for all customers handling personal data.
Notification to the customer within 72 hours of detection, with technical detail and remediation plan. Supervisory authority is notified when applicable.
Yes. The full SOC 2 Type II report is available under NDA for enterprise customers during evaluation.
Yes. For regulated sectors (healthcare, banking, defense), the decision engine and data can run on the customer's infrastructure. The conversational layer remains managed.
SECURITY CONTACT
If your DPO needs to talk, if you find a vulnerability or if you want to review the SOC 2 report under NDA, write to us.
In 30 minutes we'll show you how Twinny connects to your operation and starts delivering results from week 1.