1. Who we are
This Privacy Policy is issued by Onboard AI ("Onboard AI", "we", "us", "our"), originating from India.
⚠️ Placeholder entity: Onboard AI is not yet an incorporated legal entity as of this draft. This document cannot be published or made binding until incorporation is complete.
Contact: gmpaliwal21@gmail.com
EU Representative (GDPR Art. 27): Not currently applicable. Onboard AI does not currently serve EU-based customers. We will appoint an EU representative before processing the personal data of individuals in the EU.
India DPDP grievance contact: gmpaliwal21@gmail.com (Grievance Officer) — the Digital Personal Data Protection Act, 2023 requires a named grievance-redressal contact. Confirm final designation once incorporated.
2. Scope and roles: controller vs. processor
Onboard AI is a B2B software product. Our direct customers are businesses ("workspaces") who use the product to manage their own clients. This creates three parties:
- Onboard AI — the service provider
- The workspace — our paying customer (a business)
- The workspace's end-clients — the workspace's own customers, whose data may appear in the product because the workspace connected its own tools (Slack, Gmail, CRM, calendars) to Onboard AI
Our role depends on the type of data:
| Data category | Our role | Why |
|---|---|---|
| Workspace staff account data (names, emails, roles of people who work at the workspace) | Mixed — controller for our own operational purposes (account security, verification emails, audit logging, usage metering, billing); processor-like for day-to-day membership administration, which the workspace controls | We collect the first admin's details directly at sign-up and set our own security/metering purposes; the workspace decides who has an account |
| End-client data (client records, contacts, notes, health scores) entered or synced into the product | Processor | The workspace decides what is entered and which clients exist; we process it to provide the service |
| Business communications pulled from connected tools (Slack, Gmail, CRM events, replies) | Processor | The workspace chooses what to connect; we classify and store this data to provide the service, not for our own purposes |
| Meeting recordings and transcripts | Processor, with some Onboard AI-set defaults (retention period, consent notice wording) | The workspace starts and owns the recording; data subjects may include people outside the workspace (meeting participants) who have no direct relationship with Onboard AI |
| Operational and AI logs (AI usage logs, system logs) | Controller | Collected for our own purposes — billing, plan limits, service operation, security |
Where we act as a processor, your workspace (our customer) is the data controller for your own end-clients' and contacts' personal data. If you are a workspace administrator, you are responsible for having a lawful basis to process your end-clients' data through Onboard AI, and for responding to their data-subject requests (we will assist as described in Section 8).
3. What we collect
3.1 Account and workspace data
Name, email, role, company profile, billing contact, security contact, login activity, and audit trail entries for actions taken in the product.
3.2 Data from connected tools (processor role)
When a workspace admin connects an integration, we pull and store business data from that tool, including but not limited to: Slack messages/files/reactions, Gmail messages, Google Calendar events, CRM records (HubSpot, Salesforce, Pipedrive), issue-tracker data (Jira, Linear, GitHub, GitLab), and document content (Notion, Confluence). This data is encrypted at rest (AES-256-GCM) for sensitive fields.
3.3 Meeting data
With Zoom, Microsoft Teams, or Google Meet integrations connected, we may record, transcribe, and summarize meetings. Recordings and transcripts are stored encrypted. Data subjects include meeting participants who may not be Onboard AI account holders.
Meeting consent: For Google Meet, our recording bot posts an in-meeting notice that the meeting is being recorded and transcribed. For Zoom and Teams, Onboard AI relies on the native recording notice provided by those platforms; it is the workspace's responsibility to ensure meeting participants are informed in line with applicable law.
3.4 AI processing data
Content sent to our AI providers for classification, summarization, and the AI Companion assistant, including a redacted version of event text, transcripts, and messages (see Section 6 on subprocessors for what this data flow involves).
3.5 Operational data
AI usage logs (provider, model, token counts, feature used — for billing and plan-limit purposes), audit logs (~90 action types), and system/application logs.
4. How we use data
- To provide and operate the product (classification, health scoring, automation, AI Companion, reporting)
- To secure accounts (login, audit trail, rate limiting)
- To meter usage against plan limits and, where applicable, billing
- To communicate service-related notices (verification emails, security alerts)
- We do not use workspace or end-client data for advertising, and we do not sell personal data
5. Retention
| Data type | Default retention | Configurable by workspace admin? |
|---|---|---|
| Business communications / events | 365 days | Yes, 30–730 days. A shortened limit takes effect 7 days after saving. |
| AI Companion chat history | 180 days | Yes, 30–730 days, same 7-day change window |
| Meeting recordings and transcripts | 90 days | Yes, 30–730 days, same 7-day change window |
| AI training examples (used to improve classification accuracy for your workspace) | 180 days, except examples that were manually corrected by a workspace user, which are retained until the workspace is deleted | No — platform-wide fixed policy |
| Audit logs, AI usage logs | Retained for the life of the workspace | No |
| Account data | Retained for the life of the workspace | N/A |
When a retention period is shortened, the previous (longer) period continues to apply for 7 days before the shorter period takes effect, to give admins a window to reconsider.
6. Subprocessors
We use the following subprocessors to provide the service. All receive only the data necessary for their function.
| Subprocessor | Purpose | What it receives | Region |
|---|---|---|---|
| Railway | Application hosting, database, cache | All application data in processing | us-west2 (United States) |
| Ollama Cloud | AI model inference (classification, summarization, AI Companion) | Redacted text content sent for AI processing | United States. Prompts and responses are processed transiently and not stored beyond the request (per Ollama's privacy policy). |
| Pinecone | Vector embeddings for search and AI retrieval | Raw (unredacted) text used to generate embeddings, plus vectors and limited metadata | AWS us-east-1 (unconfirmed — verify in Pinecone console before publishing) |
| Resend | Transactional email (account verification codes) | Recipient email address, workspace name | United States |
Note on Pinecone: unlike our AI completion calls, text sent to Pinecone for embedding generation is not run through our PII redaction step first. This is disclosed here for transparency; workspace admins should be aware that client communications, transcripts, and CRM text sent for embedding reach Pinecone in raw form.
If a workspace connects its own third-party tools (Slack, Gmail, CRM, etc.), those are the workspace's own service providers, not Onboard AI subprocessors — we access them only as directed by the workspace through OAuth.
Change notice: We do not currently have an automated subprocessor-change notification system. For the pilot period, we will notify affected workspace admins by email before adding or replacing a subprocessor that processes their data.
7. Backups
We maintain daily encrypted database backups, retained for 14 days, stored separately from live data. Deleted or purged data may persist in backups for up to 14 days after deletion. We do not currently offer point-in-time recovery or a managed/redundant backup service; this is a known limitation we plan to address as the product scales beyond pilot.
8. Data subject rights and deletion
Workspace admins can request deletion of their entire workspace at any time. Deletion has a 30-day grace period (cancellable), after which we permanently purge the workspace's data, including stored files, vector embeddings, and database records, subject to the backup-retention disclosure in Section 7.
Individuals whose data appears in a workspace (e.g., an end-client's contact referenced in a Slack message) should direct data requests to the workspace that manages their relationship, since that workspace is the data controller for that data. Onboard AI will assist the workspace in fulfilling such requests where technically possible. We do not currently have an automated per-person erasure mechanism for data originating from third-party integrations; such requests are handled manually on a best-effort basis during the pilot period.
9. Security
- Field-level encryption (AES-256-GCM) for sensitive data at rest
- Multi-tenant isolation enforced at the application layer
- SSO support (Google, Microsoft)
- Audit logging of account and data actions
- Access to the encryption key is limited to Onboard AI's operations; note that this means Onboard AI technically has the capability to decrypt stored data as part of providing support and operating the service
10. International data transfers
Our infrastructure (Railway) runs in the United States (us-west2). If you or your end-clients are located in the EU, UK, or other regions with data-transfer restrictions, this means personal data is transferred to the United States for processing.
⚠️ Unconfirmed / not yet in place: a cross-border transfer mechanism (e.g., Standard Contractual Clauses) between Onboard AI and EU-connected customers has not yet been established. This must be resolved before onboarding any pilot customer whose data flows include EU data subjects.
11. Age restriction
Onboard AI is a business-to-business product intended for use by working professionals (18+) at the businesses who are our customers. It is not directed at, and we do not knowingly collect data from, individuals under 18 as end users of the product. Note that the product does process messages and communications belonging to our customers' own clients as part of providing the service (see Section 2); Onboard AI acts as a processor for that data, and the workspace (our customer) is responsible for ensuring appropriate handling of any such data relating to minors.
12. Changes to this policy
We will notify workspace admins by email of material changes to this policy and update the "Last updated" date above.
13. Contact
Questions about this policy, or to exercise applicable data rights: gmpaliwal21@gmail.com