Legal
Data Processing Addendum
What changed on November 1, 2026: This document is now called the Data Processing Addendum. Its terms are unchanged. What changed on October 31, 2026: Sections 2, 3, 4, 5, 6 and 13 and the TOMs Appendix now reflect that Intended is an authority layer only. Intended does not execute actions in your systems, run connectors, or store credentials for your third-party systems, so connector execution data and connector credentials are no longer processing categories; Section 4 adds outcome reports your agents choose to submit and your own AI provider key for optional AI-assisted classification. What changed on October 29, 2026: Section 4 now lists exactly what is stored for each decision; Section 6 makes clear that rules-only processing is the default and that Intended engages no third-party AI provider (the only third-party AI processing is through a provider key you connect yourself); Section 10 states that Intended sends no data to third-party AI providers, keeps the ban on sub-processors training on your data, and adds a product-improvement clause (Intended may use machine learning, including models it trains and hosts itself, never sent to a third-party AI provider; Tier 1 covers de-identified structural signals only, on by default for Free and Platform and off for Enterprise, with a one-switch opt-out; prompts are used only if you separately opt in); Sections 10 and 16 limit the 90-day notice to new recipients or new processing locations; Section 17 adds the Enterprise default and legal holds. Intended had no customers when these changes were made, so they take effect on this date.
1. Scope and application
This Data Processing Addendum ("DPA") supplements the Intended Terms of Service and applies when Intended, Inc. ("Intended", "Processor") processes personal data on behalf of the Customer ("Controller") in connection with the Authority Runtime Services. This DPA applies to all personal data processed through the Services, including intent metadata, authority decision records, escalation data, and account information. In the event of conflict between this DPA and the Terms of Service, this DPA prevails with respect to data protection matters.
2. Definitions
- "Personal Data" — any information relating to an identified or identifiable natural person processed through the Services
- "Processing" — any operation performed on Personal Data, including collection, storage, evaluation (risk scoring), token issuance, and deletion
- "Sub-processor" — a third-party entity engaged by Intended to process Personal Data on behalf of the Controller
- "Authority Decision Data" — intent metadata, risk evaluations, policy outcomes, token claims, and audit records generated during authority evaluation
- "Security Incident" — a breach of security leading to accidental or unlawful destruction, loss, alteration, or unauthorized disclosure of Personal Data
- "Standard Contractual Clauses" or "SCCs" — the standard contractual clauses approved by the European Commission for the transfer of personal data to third countries
- "UK Addendum" — the International Data Transfer Addendum to the EU Standard Contractual Clauses issued by the UK Information Commissioner's Office
3. Processor obligations
- Process Personal Data only on documented instructions from the Controller (as configured through policies and platform settings)
- Ensure that persons authorized to process Personal Data have committed to confidentiality obligations
- Implement and maintain technical and organizational security measures as described in Section 5 and the TOMs Appendix
- Engage sub-processors only with prior notice and under equivalent data protection obligations, in accordance with Section 6
- Assist the Controller in responding to data subject requests (access, correction, deletion, portability) within the timeframes required by applicable law
- Assist the Controller in ensuring compliance with security, breach notification, and data protection impact assessment obligations
- Cooperate with the Controller in conducting data protection impact assessments (DPIAs) where required under GDPR Article 35
- Delete or return all Personal Data upon termination of Services, subject to legal retention requirements
- Make available all information necessary to demonstrate compliance and allow for audits
- Maintain a record of processing activities carried out on behalf of the Controller in accordance with GDPR Article 30(2)
4. Categories of data processed
Account data
- Names and email addresses of account holders and team members
- Company name, role, and account configuration
Authentication data (Operator Console via Auth0 / Okta, Inc.)
- Email address used to authenticate to the Operator Console at login.intended.so
- Passkey / WebAuthn credential identifiers (the Console uses passkey-primary authentication)
- Account-linking metadata (when a user links Google and password identities for the same email)
Authority decision data
- Intent parameters: action type, target system, environment, agent identifier
- Risk evaluation: 8-factor risk scores, tier classification, contributing factors
- Policy outcomes: matched rules, thresholds, escalation triggers
- Token claims: action scope, target, expiry, nonce, signing key reference
- Audit entries: hash-chained records with SHA-256 integrity, timestamps, and decision outcomes
- Escalation records: approver identity, approval/denial status, rationale text
- Action descriptions and metadata as submitted, except that tool-call argument values are stored only as a SHA-256 digest and length and are removed from action descriptions before storage
- Natural-language requests (optional intake): a digest of the request text; the structured classification result; and, only if the Controller enables raw prompt logging, a pattern-redacted copy encrypted under a tenant-specific key
- Decision provenance: the classifier, classifier version, knowledge-pack version and rule-set version (with a hash of the exact rules evaluated) behind each decision, recorded on every decision path and shown in its receipt
Outcome reports and integration data
- Outcome reports the Controller's agents choose to submit after acting on a decision (outcome status and related metadata as submitted)
- Runtime integration artifacts generated by Intended for customer-operated third-party runtimes, including policy bundles and preset or endpoint configuration metadata
- Customer-supplied AI provider API keys, only if the Controller turns on optional AI-assisted classification (encrypted with AES-256-GCM per tenant, never logged or exported)
Data Intended does not process
- Intended does not execute actions in the Controller's systems or in third-party systems, and does not collect or store the Controller's credentials for those systems. The Controller's own agents perform actions after receiving a decision and, where issued, a signed authority token. The results of those actions are not sent to Intended unless the Controller reports an outcome
5. Security measures
- Encryption at rest: AES-256-GCM for all sensitive data and signing keys
- Encryption in transit: TLS 1.3 for all API, database, and cache connections
- Per-tenant isolation: Database-scoped queries (application + Postgres row-level security), separate Ed25519 signing key pairs
- Access control: Role-based access (4 runtime roles, 24 permissions) enforced at middleware level
- Audit integrity: SHA-256 hash-chained ledger with tamper detection
- Key management: Automated key rotation lifecycle (ACTIVE → PREVIOUS → RETIRED)
- Nonce protection: Single-use nonces consumed on first verification
- Infrastructure: AWS with VPC isolation, security groups, and encrypted storage volumes
- Monitoring: CloudWatch alarms, CloudTrail logging, and automated anomaly detection
- Fail-closed architecture: Unresolvable evaluations result in denial, not bypass
6. Sub-processors
Intended maintains a current, authoritative list of sub-processors at intended.so/legal/subprocessors. The Controller will be notified at least thirty (30) days in advance of any new sub-processor engagement or material change to an existing sub-processor. The Controller may object to a new sub-processor within twenty (20) days of notification by providing written notice to dpa@intended.so with a reasonable basis for the objection. If the Controller objects and Intended cannot reasonably accommodate the objection, either party may terminate the affected Services upon thirty (30) days written notice, with a pro-rata refund of prepaid fees for the affected Services. Enterprise customers may negotiate a longer advance-notice period (e.g., sixty (60) days) in their Order Form or enterprise agreement, which will take precedence over this standard period.
Current sub-processors
- The full, authoritative list of active and conditional sub-processors is published at intended.so/legal/subprocessors and is incorporated into this DPA by reference.
- Active sub-processors: Amazon Web Services (AWS) — infrastructure hosting, compute, storage, key management, and database services (US regions); Stripe, Inc. — payment processing and subscription billing; Resend, Inc. — transactional email delivery; Twilio Inc. — SMS notifications (engaged only when Customer enables SMS-based features); Auth0 (operated by Okta, Inc.) — Operator Console authentication (login.intended.so), identity, account-linking, and passkey/WebAuthn credential management (United States).
- AI providers: Intended engages no third-party AI provider as a sub-processor. AI policy drafting and product improvement run on models Intended hosts in its own infrastructure. Rules-only processing is the default. If Customer turns on AI-assisted classification and connects its own OpenAI or Anthropic API key, that provider processes those requests at Customer's direction and under Customer's own agreement with it.
Customer-managed third-party systems
- Systems the Controller's agents act on and customer-operated runtimes, including GitHub, Atlassian, ServiceNow, and NVIDIA OpenShell / NVIDIA NemoClaw, are not Intended sub-processors. Intended does not call or send Personal Data to them, and generating a policy artifact for a customer-operated runtime does not make it a sub-processor
- Processing performed directly in those systems is governed by the Customer's direct relationship with the applicable third party and the Customer's own instructions and credentials
7. Security incident notification
Intended will notify the Controller of a confirmed Security Incident in accordance with the following timelines:
- GDPR: Notification without undue delay, and no later than 72 hours after becoming aware of a breach likely to result in a risk to the rights and freedoms of natural persons
- Other jurisdictions: Notification without undue delay, and in compliance with applicable breach notification laws
- Notification will include: nature of the incident, categories and approximate number of data subjects affected, likely consequences, and measures taken or proposed
- Intended will cooperate with the Controller's investigation and provide ongoing updates until the incident is resolved
- Intended maintains an incident response plan with defined roles, communication procedures, and remediation protocols
- Intended will document all Security Incidents, including those that do not trigger notification obligations, and make records available to the Controller upon request
8. International data transfers
Intended processes data in the United States. For transfers of personal data from jurisdictions with transfer restrictions, Intended provides the following safeguards:
- European Economic Area (EEA): Intended relies on the Standard Contractual Clauses (SCCs) as approved by the European Commission (Implementing Decision (EU) 2021/914) for transfers of personal data from the EEA to the United States. The SCCs are incorporated by reference into this DPA as Module Two (Controller to Processor)
- United Kingdom: For transfers of personal data from the UK, Intended relies on the UK International Data Transfer Addendum (UK IDTA), Version B1.0, Module Two (Controller to Processor), as issued by the UK Information Commissioner's Office under Section 119A of the UK Data Protection Act 2018. The UK IDTA supplements the SCCs with UK-specific safeguards and is incorporated by reference into this DPA.
- Switzerland: Intended relies on the SCCs as recognized under the Swiss Federal Act on Data Protection (FADP / nDSG), with the modifications required by the Swiss Federal Data Protection and Information Commissioner (FDPIC) for transfers from Switzerland
- Supplementary measures include encryption at rest and in transit, per-tenant isolation, access controls, and fail-closed architecture as described in Section 5
- Enterprise customers requiring data residency in specific regions should contact Intended to discuss deployment options
9. Schrems II supplementary measures
In addition to Standard Contractual Clauses (SCCs) and UK IDTA, Intended implements the following supplementary technical and organizational measures to address post-Schrems II requirements:
- Encryption and Key Management: Personal data encrypted at rest (AES-256-GCM) and in transit (TLS 1.3). Encryption keys managed per-tenant. Master keys derived from tenant-controlled key material where technically feasible. US government authorities cannot decrypt data without Intended's cooperation.
- Data Isolation: Per-tenant isolation prevents aggregation of EU data with other customer data, limiting exposure in government requests.
- Fail-Closed Architecture: Unresolvable authority evaluations result in denial, not approval, preventing over-inclusive government access.
- Government Access Notification: Intended will notify Controller of any legal demand (subpoena, FISA order, NSL) requiring disclosure of personal data, unless Intended is legally prohibited. Notification will be sent to the Controller's designated contact without undue delay after receipt, and within any shorter period specified in the applicable Order Form, unless Intended is legally prohibited from doing so.
- Transparency Reporting: Intended publishes annual transparency report detailing number of government access requests, jurisdiction, and Controller's response (aggregate, non-identifying).
- Objection and Termination Rights: Controller may object to processing upon government access demand. If Intended cannot accommodate objection, either party may terminate affected Services without penalty.
- These measures do not eliminate risk of compelled US government access but substantially reduce risk and provide Controller with remedies and visibility.
10. Model training and product improvement
Intended does not send Personal Data or Authority Decision Data to any third-party AI provider, and sub-processors shall NOT use Personal Data or Authority Decision Data to train, fine-tune, or improve their models (Section 12). Where the Controller connects its own AI provider key, that provider's use of the data is governed by the Controller's own agreement with it. Intended uses Personal Data and Authority Decision Data to train or improve its own models, rules or systems only as described under Product improvement below, or with the Controller's explicit prior written consent.
By default, authority decisions for the Controller's agents are made by deterministic rules, the Controller's policies and its people; no AI model makes, scores or influences them unless the Controller turns on an optional AI feature. Optional AI features are disabled by default and are enabled only by the Controller (Settings → Privacy & AI). The Controller may block all such processing for its tenant at any time.
Product improvement
- Tier 0: Intended may use the Controller's own records to improve how the Services work for the Controller's tenant. This stays within the Controller's tenant and is part of providing the Services.
- Tier 1: the Controller authorizes Intended to derive de-identified structural signals from Authority Decision Data (Open Intent codes, decisions and outcomes, argument names, timing, and tool and target identifiers), with Controller, user and agent identifiers removed, and to use them with machine learning, including training Intended's own models, to improve classification accuracy and build product intelligence. Tool-call argument values are never stored and so are never used. Once de-identified, the signals are not Personal Data. This authorization is on by default for Free and Platform plans and off by default for Enterprise plans. The Controller may withdraw or give it at any time with one switch in Settings → Privacy & AI; enabling "Block all AI" also withdraws it. Withdrawal stops contribution from then on.
- Prompts: natural-language prompts are never used for product improvement unless the Controller separately enables "Include redacted prompts in product improvement" in Settings → Privacy & AI. That setting is off by default on every plan and can be enabled only while raw prompt logging is on and "Block all AI" is off. When enabled, prompts are redacted before any use (personal data and secrets such as email addresses, phone numbers, URLs, keys and tokens are removed) and are processed only on models Intended hosts in its own infrastructure.
- Tier 2: any use beyond Tier 1 requires a separate written agreement with the Controller.
- Product improvement runs on models Intended hosts in its own infrastructure. Intended does not send Personal Data or Authority Decision Data to any third-party AI provider (including OpenAI or Anthropic) for product improvement, and no AI sub-processor is engaged for it.
- Improvements reach the default decision path only as new, versioned deterministic rule sets, and every decision receipt shows which classifier and rule-set version made the decision.
- Intended owns the resulting models, rules and de-identified aggregate data. The Controller owns its authority decision history.
- Intended will give at least 90 days' advance written notice, with a right to object under Section 6, before any change that sends Personal Data or Authority Decision Data to a new recipient (a new AI provider or any new sub-processor) or changes where it is processed or transferred. Improvements to models and pipelines that Intended hosts itself, operating within the Tier 0, Tier 1 and Tier 2 contribution settings above, are expressly exempt.
11. Data protection impact assessments
Intended will provide reasonable cooperation and assistance to the Controller in conducting data protection impact assessments (DPIAs) where required under GDPR Article 35 or equivalent provisions of other applicable data protection laws. Intended will provide information about its processing activities, technical and organizational measures, and sub-processors as needed for the Controller to complete its assessment. Requests for DPIA cooperation should be directed to dpa@intended.so.
12. Sub-processor obligations
Each sub-processor engaged by Intended shall implement obligations substantially equivalent to this DPA, including but not limited to:
- Process Personal Data only on documented instructions from Intended (which, in turn, follows Controller instructions).
- Implement and maintain technical and organizational security measures per Section 5 (Appendix TOMs).
- Notify Intended of any Security Incident within 24 hours of discovery.
- Assist with data subject access, deletion, correction, and portability requests within timeframes specified in Section 13.
- Delete or return Personal Data upon termination, except where legal retention applies.
- Submit to audits and provide compliance documentation to Intended (which Intended may share with Controller under NDA).
- Not engage further sub-processors without prior written approval from Intended.
- Implement the same prohibition on model training per Section 10.
- Intended shall ensure all sub-processor agreements are executed before engagement and shall provide evidence of compliance upon request.
13. Data subject rights assistance
Assist the Controller in responding to data subject requests in accordance with the following timelines and procedures:
- Requests for Access (GDPR Art. 15, CCPA §1798.100): Extract all personal data held for the data subject in machine-readable format (CSV, JSON) within 10 business days of request. Provide all authority decisions, audit records, outcome reports, and account data.
- Requests for Deletion (GDPR Art. 17, CCPA §1798.105): Delete all personal data associated with data subject within 10 business days, except where legal retention (SOX 7-year, GDPR Art. 17(3)) or ongoing disputes require retention. Notify Controller if deletion is not fully possible and explain legal basis for retention.
- Requests for Correction (GDPR Art. 16, CCPA §1798.106): Correct factual inaccuracies in account data (name, email) within 5 business days. For authority decision data, correction may not be technically possible (hash-chained, tamper-evident audit ledger); in that case, Intended will flag the record as disputed and provide attestation of the error.
- Requests for Portability (GDPR Art. 20): Provide all personal data and authority decision records in structured, machine-readable format (CSV, JSON, XML) within 10 business days, suitable for import to another processor.
- Requests for Objection (GDPR Art. 21): Upon Controller's instruction that data subject has objected to processing, Intended will cease processing that data subject's data within 5 business days, except where legal basis for continued processing applies.
- Cost: Routine requests (up to 5 per calendar year per data subject) are included in Services fees. Requests in excess of 5 per data subject per year are subject to professional services fees at Intended's standard rate as specified in the applicable Order Form or quoted on request; contact dpa@intended.so for the current rate.
- Procedure: Data subject rights requests should be submitted to dpa@intended.so with sufficient information to identify the data subject and the requested action. Intended will acknowledge receipt within 2 business days and provide progress updates every 5 business days if response exceeds 10 business days.
14. Limitation of liability for DPA breaches
Intended's liability for DPA breach is governed by the Terms of Service Section 14 (Limitation of Liability) and the liability cap specified in any MSA. Specifically:
- Intended's aggregate liability for data protection claims (breach of confidentiality, failure to implement security measures per §5, unauthorized access) shall not exceed the liability cap set out in the Terms of Service Section 14 or the amount specified in the applicable MSA or Order Form, whichever is lower.
- Intended is not liable for losses arising from Controller's use of personal data or authority decisions (e.g., customer's own discriminatory use of Intended's risk scores) if Intended has complied with §3 (instructions) and §5 (security measures).
- Intended is not liable for sub-processor breaches unless Intended failed to exercise reasonable oversight of the sub-processor or failed to flow down equivalent DPA obligations per Section 6.
- Carve-outs from liability cap: Gross negligence, willful misconduct, breach of confidentiality (Section 3), IP indemnification (Terms §16A), and Intended's payment obligations.
16. AI Act compliance confirmation
Intended confirms the following regarding Personal Data and AI processing:
- Personal Data and Authority Decision Data are used to train or improve AI systems only as described in Section 10 (de-identified product improvement, subject to the Controller's contribution settings) or with explicit prior written Controller consent.
- Intended complies with EU AI Act (Regulation 2024/1689), Articles 4–7 (prohibited practices), Articles 8–15 (high-risk AI), and Articles 51–55 (GPAI deployer obligations), to the extent Personal Data is processed in connection with AI.
- Intended maintains documentation of processing and AI safeguards required by AI Act Article 13(3) and Article 5(2)(a) and will provide to Controller upon request.
- Intended will give at least 90 days' advance notice, with a right to object under Section 6, before any change that sends Personal Data or Authority Decision Data to a new recipient (a new AI provider or any new sub-processor) or changes where it is processed or transferred. Improvements to models and pipelines that Intended hosts itself, operating within the published Tier 0, Tier 1 and Tier 2 contribution settings (Section 10), are expressly exempt.
- Controller remains responsible for compliance with AI Act Articles 6(1)–(2) (high-risk AI assessment and conformity) and Article 22 (GDPR right to human review when AI makes decisions about natural persons).
17. Data retention and deletion
- Authority decision data is retained according to the Controller's plan tier (Free 30 days, Platform 1 year, Enterprise custom up to 7 years; 7 years if no custom window is set), enforced by a daily deletion job
- If a legal hold is active for the Controller's tenant, no data is deleted while the hold is in place
- Account data is retained for the duration of the active agreement plus 30 days for export
- Upon termination, Intended will delete or return Personal Data within 30 days, except where retention is required by law
- Audit records subject to regulatory retention requirements (e.g., SOX 7-year) are retained on encrypted, versioned storage with deletion protection for the applicable retention period; immutable WORM storage (S3 Object Lock, Compliance mode), which blocks deletion before the retention period expires, is on the enterprise roadmap and available for dedicated deployments
- The Controller may request certification of deletion upon completion of the deletion process
18. Audits and compliance
The Controller may audit Intended's compliance with this DPA once per year with 30 days written notice. Audits will be conducted during business hours and will not unreasonably interfere with operations. Intended has scoped an independent third-party penetration test and is engaging a qualified security firm; independent testing runs on at least an annual cadence thereafter. Intended will provide available compliance documentation upon request under NDA, including SOC 2 reports and penetration-test summaries once the respective audit and first assessment are complete. Enterprise customers may request additional compliance documentation through their account team. Intended will maintain a record of processing activities in accordance with GDPR Article 30(2) and make it available to the Controller and supervisory authorities upon request.
19. Duration and termination
This DPA remains in effect for the duration of the Services agreement. Obligations regarding data deletion, retention, confidentiality, and cooperation with data protection authorities survive termination. In the event of conflict between this DPA and the Terms of Service, this DPA prevails with respect to data protection matters.
20. Data processing jurisdiction
- Primary processing location: United States (AWS us-east-1)
- Data may be processed in AWS regions selected by Customer for Enterprise deployments
- International transfers are governed by Standard Contractual Clauses (SCCs) for EEA/UK transfers, as described in Section 8
- Sub-processor list is maintained and updated with thirty (30) days advance notice in accordance with Section 6
21. Contact
For DPA-related inquiries, data protection officer contact, or to request a signed copy: dpa@intended.so or legal@intended.so. For EU-specific data protection inquiries: eu-privacy@intended.so.
Appendix: Technical and organizational measures (TOMs)
The following technical and organizational measures are implemented by Intended to protect personal data processed under this DPA:
Access control
- Role-based access control with four runtime roles and twenty-four granular permissions
- Multi-factor authentication (MFA) support for all account holders
- API key authentication with hashed storage and prefix-based identification
- Brute-force protection with rate limiting on authentication endpoints
- Automated session expiry and re-authentication requirements
Encryption
- AES-256-GCM encryption at rest for all sensitive data, signing keys, and customer-supplied AI provider API keys
- TLS 1.3 encryption in transit for all connections
- Ed25519 key pairs (RSA-4096 selectable) for authority token signing (per-tenant isolation)
- HMAC-SHA-256 for evidence bundle integrity verification
- HKDF key derivation for derived key material
Data integrity
- SHA-256 hash-chained audit ledger with tamper detection
- Single-use nonces to prevent replay attacks
- Cryptographic receipts for all authority decisions
- Hash-chained, tamper-evident audit records (payload content is never altered in normal operation; hash pointers may be re-linked only by an authorized, logged remediation; database-level append-only is enforced (UPDATE/DELETE revoked from the serving role))
Infrastructure security
- AWS cloud infrastructure with VPC isolation
- Encrypted storage volumes (EBS, S3)
- Security groups and network access control lists
- CloudWatch monitoring, CloudTrail audit logging, and automated alerting
- Regular security patching and vulnerability management
Organizational measures
- Confidentiality agreements for all personnel with access to personal data
- Security awareness training for engineering and operations staff
- Documented incident response plan with defined roles and procedures
- Independent third-party penetration test scoped and being commissioned, with independent testing on at least an annual cadence thereafter
- Regular review and update of security policies and procedures