Legal
Data Residency & Retention Policy
This Data Residency & Retention Policy ("Policy") supplements the Adaptive AI Terms of Service v1.0 ("ToS"), Privacy Policy v1.1 ("Privacy Policy"), and Data Processing Agreement v1.0 ("DPA"). It describes where Customer Data is stored, how long it is retained, and how deletion is handled.
In the event of conflict: (1) the Order Form controls; (2) then the DPA; (3) then the ToS; (4) then this Policy; (5) then the Privacy Policy (ToS §13.2). All capitalized terms not defined herein have the meanings given in the ToS.
1. Data Residency
1.1 Default Residency: United States
By default, all Customer Data is stored and processed in the United States. Adaptive AI's primary infrastructure providers and their default regions are:
| Provider | Service | Default Region | Data Types |
|---|---|---|---|
| Railway | Application hosting (single-tenant) | US regions | AI Worker compute, application logic, session state |
| Supabase | Authentication and database (PostgreSQL) | US East | Account data, conversation transcripts, configuration, session metadata |
| Cloudflare R2 | Object storage | US | Knowledge base files, voice recordings, uploaded documents |
| Cloudflare Vectorize | Vector database | US | Embedding vectors for knowledge base search |
| LiveKit Cloud | Real-time communications (WebRTC) | US regions | Voice and video media streams (transient; not stored by LiveKit) |
1.2 EU Data Residency (Available on Request)
For Customers who require data residency within the European Economic Area ("EEA"), Adaptive AI offers EU-resident deployment on request. EU residency is available on qualifying Order Forms and includes:
| Provider | EU Region | Notes |
|---|---|---|
| Railway | EU regions | Dedicated single-tenant compute in EU |
| Supabase | EU (Frankfurt) | Dedicated PostgreSQL instance in EU |
| Cloudflare R2 | EU | Object storage with EU jurisdiction hint |
| Cloudflare Vectorize | EU | Vector storage in EU region |
| LiveKit Cloud | EU (Frankfurt) | Real-time media processed in EU |
When EU data residency is specified in an Order Form, Adaptive AI will not store or process Customer Data outside the EEA without Customer's prior written consent, except: (a) where required by Applicable Data Protection Law; or (b) for transient processing that is covered by the Standard Contractual Clauses in the DPA (DPA §9; DPA Annex C).
1.3 UK Data Residency
UK data residency is available under the same terms as EU data residency, using UK-based or EEA-based infrastructure. The UK International Data Transfer Addendum to the EU SCCs applies (DPA Annex C).
1.4 Custom Residency Requirements
For Customers with residency requirements beyond US and EU (e.g., Australia, Canada, Japan), Adaptive AI will evaluate infrastructure availability and provide a custom residency addendum to the Order Form. Contact admin@adaptiveap.com.
1.5 Data Residency for AI Model Inference
When AI Workers process conversations, inference requests are sent to large language model (LLM) providers. By default, inference requests may be processed outside the Customer's configured data residency region.
BYOK (Bring Your Own Keys): Customers who require inference within a specific region may use their own API keys to route requests through their own provider agreements, giving them direct contractual and jurisdictional control over inference processing (ToS §5.7; Privacy Policy §10.2).
Adaptive AI configures all LLM providers to prohibit training on submitted data, regardless of whether BYOK is used (ToS §5.2).
2. Sub-processors and Data Flow
2.1 Sub-processor List
A current list of Sub-processors, including their locations and processing activities, is maintained at legal.adaptive.ai/subprocessors.
2.2 Sub-processor Changes
Adaptive AI will notify Customer account administrators via email at least 30 days before adding any new Sub-processor. Customers may object to new Sub-processors under the terms of DPA §6.4 and ToS §5.6.
2.3 Data Flow Summary
End User ──► [LiveKit Cloud / Channel Provider] ──► [Railway: AI Worker Compute]
│
├──► [LLM Provider: inference]
├──► [Supabase: transcripts, config, metadata]
├──► [Cloudflare R2: recordings, files]
└──► [Cloudflare Vectorize: knowledge vectors] All data flows are encrypted in transit (TLS 1.2+). All stored data is encrypted at rest (AES-256). Single-tenant isolation applies at every layer (ToS §5.10; Privacy Policy §4; DPA §5.2).
3. Retention Schedule
3.1 Default Retention Periods
Adaptive AI retains data only as long as necessary to provide the Platform, meet legal obligations, or as configured by Customer (Privacy Policy §9; ToS §5.8).
| Data Type | Default Retention | Configurable | Minimum | Maximum | Legal Basis for Retention |
|---|---|---|---|---|---|
| Account and billing records | Contract term + 7 years | No | N/A | N/A | Tax and financial compliance (26 USC §6501; IRS Rev. Proc. 98-25) |
| Conversation transcripts | 90 days | Yes (per Order Form) | 7 days | 365 days | Contractual necessity; Customer configuration |
| Voice recordings | 30 days | Yes (per Order Form) | 7 days | 365 days | Contractual necessity; Customer configuration |
| Session telemetry and logs | 30 days | No | N/A | N/A | Security and performance diagnostics; legitimate interest |
| Support tickets | 3 years | No | N/A | N/A | Legitimate interest; dispute resolution |
| Website analytics | 13 months | No | N/A | N/A | Legitimate interest (GDPR Art. 6(1)(f)) |
| Security and audit logs | 1 year | No | N/A | N/A | Security, legal compliance, and audit requirements |
| Deletion audit logs | 1 year | No | N/A | N/A | Compliance and audit requirements (DPA §10) |
| PHI (when BAA in effect) | 6 years | Yes (up to 10 years) | 6 years | 10 years | HIPAA (45 CFR §164.530(j)); see BAA §6 |
| BYOK API credentials | Active subscription term | No | N/A | N/A | Contractual necessity; deleted within 7 days of key removal |
3.2 Custom Retention Periods
Customers may request shorter retention periods for conversation transcripts and voice recordings via their Order Form or by contacting admin@adaptiveap.com. Custom retention periods must be at least 7 days (to support operational diagnostics and dispute resolution).
Customers who require longer retention periods (e.g., for regulatory compliance in financial services or healthcare) may request extensions via their Order Form, up to the maximum periods listed above.
3.3 Legal Holds
Adaptive AI may be required to preserve data beyond the configured retention period in response to a valid legal hold, subpoena, or court order. Where legally permitted, Adaptive AI will notify Customer of any such preservation requirement (Privacy Policy §10.3).
4. Automated Retention Enforcement
4.1 Automated Deletion
Retention periods are enforced automatically. When data reaches the end of its configured retention period, it is queued for deletion and permanently removed within 7 days of the retention expiry date.
4.2 Deletion Method
- Database records (Supabase/PostgreSQL): Hard delete from all tables; no soft-delete or tombstone retention beyond the 7-day processing window.
- Object storage (Cloudflare R2): Objects permanently deleted; bucket versioning is not enabled for Customer Data.
- Vector storage (Cloudflare Vectorize): Embedding vectors permanently deleted.
- Backups: Customer Data is purged from backup systems within 30 days of deletion from primary storage.
- Log systems: Retained per the security and audit log retention period (1 year), after which they are permanently deleted.
4.3 Deletion Verification
Adaptive AI maintains deletion audit logs confirming the date and scope of each automated deletion operation. These logs are retained for 1 year and are available for Customer audit upon request (DPA §10).
5. Customer Self-Service Deletion
5.1 Platform UI
Customers may delete the following data types at any time through the Platform's account settings:
- Individual conversation transcripts and voice recordings
- Knowledge base content and files
- AI Worker configurations
- BYOK API keys
5.2 API
Customers may programmatically delete data using the Adaptive AI API. The following endpoints support deletion:
DELETE /api/v1/conversations/{id} — Delete a specific conversation and associated transcript/recordingDELETE /api/v1/knowledge/{id} — Delete a knowledge base entryDELETE /api/v1/workers/{id} — Delete an AI Worker and its configurationPOST /api/v1/account/delete-all-data — Request deletion of all Customer Data (requires confirmation)API documentation is available at docs.adaptiveap.com.
5.3 Bulk Deletion on Termination
- Customer Data remains available for export for 30 days (ToS §10.5(d); Privacy Policy §9; DPA §11.2).
- Export requests must be submitted within this 30-day window.
- After the export window closes — or upon Customer's earlier written request — Adaptive AI permanently deletes Customer Data within 30 days.
- Account and billing records are exempt from this deletion to the extent required by law (retained for contract term + 7 years).
- Upon request, Adaptive AI will provide written certification of deletion (DPA §11.3).
5.4 Right to Erasure (GDPR Article 17)
For requests related to individual Data Subject erasure rights under the GDPR, contact admin@adaptiveap.com. Adaptive AI will assist Customer in fulfilling erasure requests within the timeframes required by Applicable Data Protection Law (DPA §7; Privacy Policy §13).
6. Data Export
6.1 Export Formats
Customer Data is exportable in the following formats:
| Data Type | Export Format |
|---|---|
| Conversation transcripts | JSON, CSV |
| Voice recordings | WAV, MP3 |
| Session metadata | JSON, CSV |
| AI Worker configurations | JSON |
| Knowledge base content | Original file format |
| Analytics and telemetry | JSON, CSV |
6.2 Export Methods
- Platform UI: Export individual records or bulk export via account settings.
- API: Programmatic export via
GET /api/v1/export/{data_type}endpoints. - Termination export: Full data export available for 30 days post-termination (ToS §10.5(d)).
6.3 Data Portability (GDPR Article 20)
Where Applicable Data Protection Law grants Data Subjects the right to data portability, Adaptive AI will provide Personal Data in a structured, commonly used, machine-readable format (JSON) (Privacy Policy §13; DPA §7).
7. Healthcare Data (HIPAA)
7.1 Protected Health Information (PHI)
Adaptive AI does not process Protected Health Information ("PHI") by default. Customers who deploy AI Workers in contexts where PHI may be processed must:
- Execute a Business Associate Agreement ("BAA") with Adaptive AI before processing PHI. See BAA.
- Ensure that AI Worker configurations comply with HIPAA minimum necessary standards.
- Enable appropriate technical safeguards (encryption, access controls, audit logging) as specified in the BAA.
7.2 HIPAA Retention
Where a BAA is in effect, Adaptive AI will retain PHI in accordance with HIPAA requirements (minimum 6 years for designated record sets, or as specified in the BAA), which may override the default retention periods in Section 3.
8. Changes to This Policy
Adaptive AI may update this Policy periodically. Material changes will be notified to Customer account administrators via email at least 30 days before taking effect. For changes that materially restrict Customer's rights, the amendment process in ToS §13.3 applies.