Companion← Back to Companion

Legal

Privacy Policy

Last updated: October 15, 2026

Companion (operated by ANA Studio / AnalystAI Inc.) helps organizations run frontline knowledge, check-ins, training, and operational briefings. This policy explains what data we collect, how AI models and subprocessors handle it, how long it is kept, and how workspace admins control their workspace data.

On this page

  1. Information we collect
  2. How we use information
  3. AI processing
  4. Sharing and subprocessors
  5. Data retention
  6. Workspace admin controls
  7. Security
  8. Contact

Information we collect

We collect account information (name, email, role), workspace configuration, uploaded documents and product sheets, approved team phone numbers, employee and manager messages, voice notes, training assignments and quiz responses, and operational audit logs needed to provide the service. When a workspace connects WhatsApp, Telegram, or email, we process only the messages, media, and delivery events authorized for that workspace.

How we use information

We use workspace data solely to deliver and secure the service for that workspace: answering staff and manager questions from approved documents, transcribing voice notes, generating operating briefs, drafting training courses, routing check-ins, and recording an auditable log of workspace actions. Each workspace is logically isolated so one organization's documents and conversations are never visible to another organization.

AI processing

To answer questions, extract tables and text from uploaded files, transcribe voice notes, classify question outcomes, and draft training modules, Companion sends only the relevant workspace prompt, document passages, or audio clip to commercial AI API providers (listed under Subprocessors below).

No model training on customer data: We use commercial API tiers configured with zero training permissions. Neither Companion nor our AI model providers use customer workspace documents, messages, or voice recordings to train or fine-tune foundation models. Each model call is scoped to a single workspace and logged in the provider audit trail.

Sharing and subprocessors

We do not sell workspace or personal data, and we never share one customer's workspace data with another customer. We use the following infrastructure and service subprocessors to operate Companion:

SubprocessorRoleData processed
CloudflareEdge DNS, TLS termination, DDoS protection, and R2 object storageHTTP request metadata, IP addresses, and encrypted workspace document/media assets
VercelFrontend web application hosting and edge routingWeb session requests and rendered dashboard responses
ClerkAuthentication and session management for operators and workspace usersName, email address, authentication identifiers, and sign-in audit events
OpenRouter / OpenAI / Google Gemini / PerplexityCommercial LLM inference, embeddings, document extraction, decision classification, and audio transcriptionWorkspace document excerpts, questions, check-in responses, and voice notes submitted for processing
ElevenLabsOptional conversational voice agent and speech synthesis (when enabled for a workspace)Voice call audio, transcripts, and caller phone metadata
Meta Platforms (WhatsApp Cloud API)Inbound and outbound WhatsApp message and template deliveryRecipient phone numbers, message payloads, media attachments, and delivery status webhooks
ResendTransactional email delivery (operating briefs, verification codes, and invitations)Recipient email addresses and transactional email content
PostHogProduct usage analytics (with workspace IDs and tokens masked in URLs)De-identified page navigation and feature interaction events

Data retention

Workspace data (uploaded documents, messages, voice notes, training records, and audit logs) is retained for the duration of the customer's active workspace subscription so managers can search history, verify citations, and track training progress.

When a workspace admin deletes a document or revokes a shared report, it is removed immediately from active retrieval. Upon workspace closure or a written deletion request from a workspace administrator, all workspace data is deleted from active production databases within 30 days and aged out of encrypted disaster-recovery backups within 90 days, except where longer retention is strictly required by applicable law.

Workspace admin controls

Workspace admins control which phone numbers are approved to join, which documents are approved for sharing with frontline staff, whether peer leaderboards use anonymised initials, and which channels are connected. Admins can export or delete workspace documents at any time from the dashboard or request full workspace deletion via support.

Security

We enforce tenant-scoped database queries, composite foreign-key integrity between documents and chunks, signed webhook verification, phone possession confirmation before a number becomes active, environment-scoped secrets, and append-only audit logs for sensitive actions. Please report any security concerns promptly to hello@analystai.ai.

Contact

For privacy, data-processing agreements (DPA), subprocessor updates, or deletion requests, contact hello@analystai.ai.

© 2026 Companion
PrivacyTermsContactHome