Privacy Policy

Loop — Collaborative Intelligence Platform Operated by Launch Engine, Inc.

Effective Date: March 10, 2026 | Last Updated: November 9, 2026 | Version: 2.10

Quick Summary

Loop is built with privacy-by-design principles using secure infrastructure:

Vercel for frontend hosting and edge network delivery
WorkOS for authentication, SSO, and identity management
Convex for real-time database, file storage, vector database, and RAG
AES-256-GCM encryption for sensitive data at rest
TLS 1.2+ encryption for all data in transit
We do not sell your data. We do not train AI models on your content.

1. Introduction

Launch Engine, Inc., the maker of Loop ("Launch Engine," "we," "us," or "our") is committed to protecting your privacy and ensuring the security of your data. This Privacy Policy explains how we collect, use, share, and protect information when you use Loop, our collaborative intelligence platform.

2. Information We Collect

2.1 Account Information

When you create a Loop account, we collect:

Profile Information: First name, last name, email address, profile picture
Authentication Data: OAuth tokens from identity providers via WorkOS (Google, GitHub, Microsoft, etc.)
Organization Details: Company name, role, industry, department (if applicable)
Contact Information: Phone number (optional), social profiles (optional)

2.2 Content You Create ("Customer Data")

You own and control all content you create or upload to Loop:

Files and Documents: Text, markdown, spreadsheets, presentations, images, video, audio
Templates: Handlebars templates with variable namespaces (env, user, org, file, contact, company, input, calc)
Prompts and Agents: AI prompts, agent configurations, collections, and workflow definitions
AI-Generated Content: Documents, images, and text created using AI features
Tags and Collections: Organizational taxonomy including teamspace-scoped tags
Metadata: Descriptions, quality scores, freshness metrics, usage analytics
AI Operation Logs: When you use AI features, we log prompts, responses, context used, token counts, cost, response quality scores, operational warnings, and feature flag states within your organization's workspace for quality monitoring, evaluation, and debugging
Agent Memory: AI agents may store persistent structured memory about your preferences, decisions, and interaction history within your organization's workspace. This memory is used to personalize agent behavior and is scoped to your organization. You can view and delete agent memory at any time.
Credit Transactions: When you use AI features on paid plans, we log credit usage per operation (model, token counts, cost in credits) for billing accuracy and usage reporting
Extracted Text: Content extracted from uploaded documents for search and AI processing (controllable per-file)
Vector Embeddings: Semantic representations generated for RAG search (controllable per-file)

2.3 Integration Data

When you connect third-party services:

Slack: Messages, channels, and files you choose to sync or share
Google (Drive, Docs, Gmail, Calendar): Files you pick or import from Drive, email you ask Loop to send and, where the inbox feature is available to you, email you ask Loop to read, and calendar events, all with your permission. How we handle this data is set out in §5.4
Notion: Pages, databases, and content you explicitly share
OAuth Tokens: Encrypted at rest with AES-256-GCM
Sync Preferences: Which folders, labels, channels, or content types to sync
Integration Scope: Per-teamspace integration controls (inherit from org or restrict)

2.4 Automatically Collected Data

Usage Logs: Feature usage, page views, and interaction patterns (for product improvement)
Audit Logs: Authentication events, permission changes, and content modifications, logged across a comprehensive range of event types
Observability & Telemetry: Backend performance metrics, error traces, and query latency data collected via Axiom for infrastructure monitoring and debugging
IP Addresses: May be collected by infrastructure providers (Vercel, Convex) per their respective policies
Device Information: Browser type, operating system (for compatibility)
Onboarding Website Data: During onboarding, if you provide your company's website URL, Loop may scrape publicly available information from that website (such as company description, industry, and branding) to pre-populate your organization's context. Only public-facing content is accessed.

3. How We Use Your Information

3.1 To Provide and Improve Loop

Service Delivery: File management, document generation, collaboration, teamspace scoping
AI Features: Process documents, generate content, provide intelligent suggestions (only when you invoke AI)
Search Functionality: Full-text and semantic (RAG) search, respecting file-level permissions
Real-time Sync: Keep data synchronized across devices and team members via Convex reactive queries
Template Engine: Resolve variable namespaces and compile templates with your data

3.2 For Security and Safety

Authentication: Verify identity through WorkOS AuthKit (JWT validation on every request)
Authorization: Enforce role-based access control at organization, teamspace, and file levels
Fraud Prevention: Detect and prevent unauthorized access via audit logging
Rate Limiting: Prevent abuse and ensure service stability

3.3 For Communication

Transactional Emails: Account verification, password resets, join request notifications
Notification Digests: Configurable per-channel (email, Slack, in-app) and per-category
Service Announcements: Critical updates about security, terms changes, or service disruptions

3.4 Text Messages (SMS)

If you give us your mobile number and agree to receive texts, we send account texts only: alerts, reminders and replies to your support requests. Message frequency varies with your activity. Reply STOP to opt out at any time, or HELP for help. Msg & data rates may apply.

We do not share your mobile number or your text-message consent with any third party or affiliate for their marketing or promotional purposes. We share it only with Twilio, our text-message provider (see §5.2), to deliver the messages you asked for.

4. AI and Machine Learning

4.1 AI Processing Controls

Loop gives you granular control over AI processing at three levels:

Level	Controls
User-Level	Enable/disable AI for your personal content
File-Level	Per-file toggles: `allowEmbeddings`, `allowAnalysis`, `allowGeneration`, `allowTagging`
Organization-Level	Admins set org-wide AI policies and default model

4.2 AI Data Practices

No Training on Your Data: We do not use your content to train or fine-tune AI models, and we do not enroll your content in any provider's model-training program
Provider Data Handling: AI providers process your content to fulfill your request under their standard commercial API terms, which prohibit training their models on API inputs. Some providers may retain API inputs for a limited period for safety and abuse monitoring, per their published data-retention policies (linked in §4.3)
Organization-Scoped Logs: Loop retains AI operation data (prompts, responses, token usage, cost, quality scores, and operational diagnostics) within your organization for quality monitoring, evaluation, and debugging. This data is scoped to your organization, accessible only by org admins, and deleted when your organization is deleted. Retention policies are configurable.
Beta Review (Temporary): During our beta period, authorized Loop staff may review a sample of chat conversations, agent interactions, and usage logs to identify quality issues, improve AI outputs, and debug the platform. This access is limited to authorized team members under confidentiality obligations. This review does not include content that came from Google APIs (see §5.4). We will update this policy and provide notice before discontinuing this practice as we exit beta.
Web Search: When agents perform web searches (via Google Gemini grounding, Anthropic's web search tool, or OpenAI's search capability), your query context is transmitted to the respective provider's search infrastructure. This is governed by each provider's data retention policy (linked in §4.3).
No Cross-Org Sharing: Your AI operation data is never shared with other organizations or used to train models
Immediate Cleanup: When AI permissions are revoked on a file, derived data (embeddings, extracted text, analysis) is deleted immediately
Prompt Isolation: Your prompts and content are never shared with other users or organizations

4.3 AI Providers

Provider	Usage	Data Handling
Google (Gemini)	Content generation, analysis	API terms of service (data retention policy)
Anthropic (Claude)	Content generation, agent workflows	API terms of service (data retention policy)
OpenAI	Content generation, embeddings, transcripts of recordings	API terms of service (data retention policy)
TypeSafe	Quick decisions inside AI features, such as which action, answer or search passage fits a request	API terms of service (privacy policy)

Loop configures standard models using Claude, Gemini, and OpenAI, and uses TypeSafe's models for quick decisions inside AI features. Calls to these providers may be routed through Vercel's AI Gateway. Customers can change their default models, or add their own API keys, in organization settings.

You can disable AI processing entirely at any time. Disabling removes all derived AI data from your files.

5. Data Sharing and Disclosure

5.1 We Do Not Sell Your Data

Loop never sells, rents, or trades your personal information or content. Period.

5.2 Service Providers (Subprocessors)

We share data with the following trusted service providers, each under data processing agreements:

Provider	Purpose	Data Shared
Vercel	Frontend hosting, edge functions, CDN, and routing of AI requests (AI Gateway)	Request metadata, static assets, and AI requests and responses passing through the AI Gateway, which is set to zero data retention and routes only to providers that also keep none
WorkOS	Authentication, SSO, SCIM provisioning	User identity, auth tokens, org membership
Convex	Database, file storage, vector DB, serverless backend	All customer data (encrypted at rest)
Google AI (Gemini)	AI content generation	Content sent for processing (per provider API terms; no training on your content)
Anthropic (Claude)	AI content generation	Content sent for processing (per provider API terms; no training on your content)
OpenAI	AI content generation, embeddings, transcription	Content sent for processing, and audio recordings you ask Loop to transcribe (per provider API terms; no training on your content)
TypeSafe	AI decisions inside Loop's features	The text of a request and the context the decision reads (per provider API terms; no training on your content)
Stripe	Payment processing	Payment methods, billing info (tokenized)
Resend	Email delivery, including email you send to your own recipients through Loop	Email addresses and message content
HubSpot	CRM & marketing analytics	User profile, org name, subscription tier, industry; synced on signup, profile update, org creation, member addition, and subscription change
Twilio	SMS messaging	Message content, recipient phone numbers
Gmail API	Email delivery (user-connected)	Email content, recipients (via user's own OAuth)
LiveKit	Real-time voice sessions	Audio streams and session metadata during voice chat interactions
Google Analytics	Product usage analytics	Pseudonymous usage events and anonymized IP address; configured for analytics only (Google Signals and advertising features disabled, no data-sharing with Google for advertising)
Axiom	Backend observability and telemetry	Performance metrics, error traces, query latency (no user content)

Changes to Subprocessors. We will provide at least 30 days' advance notice before adding or replacing a subprocessor that processes customer data, via email to organization administrators and/or an update to this policy. This gives you the opportunity to review the change before it takes effect.

5.3 Legal Requirements

We may disclose data when required by law, subpoena, or legal process, or to protect our rights, safety, or property.

5.4 Google User Data

Loop's use of information received from Google APIs, including Google Workspace APIs, will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

For the Google data you connect (Drive files, Gmail messages, Calendar events):

Only for the feature you asked for: We use it to provide or improve the features you use, such as importing a file you picked or sending an email you approved
Never sold, never used for ads: We do not sell it, transfer it to advertising platforms, data brokers or information resellers, or use it to serve ads
No AI training: We do not use it to train or improve AI or machine-learning models (see §4.2)
No human reading: People at Loop do not read it unless you ask us to, it is necessary for security purposes (such as investigating abuse), or we must to comply with the law. This includes the beta review in §4.2, which does not cover content that came from Google APIs
Limited transfers: We transfer it only to the service providers listed in §5.2 as needed to provide the features you use, for security purposes, or to comply with the law

6. Data Security

6.1 Technical Measures

Measure	Implementation
Encryption in Transit	TLS 1.2+ for all API and WebSocket communications
Encryption at Rest	AES-256-GCM for OAuth tokens, API keys, and secrets
API Key Security	SHA-256 hashing, cryptographic random generation, scoped permissions
Authentication	WorkOS AuthKit — JWT tokens (15-min expiry), automatic refresh, MFA support
Infrastructure	Runs on SOC 2-certified infrastructure (Convex, Vercel); Loop's own SOC 2 Type II attestation is in progress
Audit Logging	Comprehensive security event logging (authentication, permission, and content events) with user, action, timestamp, and target
Data Isolation	Multi-tenant architecture with organization and teamspace-level isolation
SCIM Provisioning	Enterprise-grade identity sync for Okta, Entra ID, JumpCloud (enterprise tier)

6.2 Access Controls

Organization Isolation: Complete data separation between organizations
Teamspace Scoping: Content (files, tags, collections) can be scoped to teamspaces within an organization
Role-Based Access: Owner > Admin > Member > Visitor permission hierarchy
File-Level Permissions: Private, organization, teamspace, community, public, and visitors visibility levels
API Key Scoping: Read/write separation with organization binding

6.3 Your Security Responsibilities

Keep your credentials secure
Enable multi-factor authentication when available
Review organization and teamspace member permissions regularly
Report suspicious activity immediately to security@loopllm.ai

7. Your Rights and Controls

7.1 Access and Portability

View Your Data: Access all your content through the Loop interface
Export Your Data: Download files in original formats; JSON data export via exportUserData feature
API Access: Use API keys to programmatically access your data

7.2 Correction and Deletion

Edit Content: Update or correct your information anytime
Delete Files: Remove individual files or bulk delete. By default deleted files go to a 30-day trash you can restore from; you can instead delete immediately, and an Enterprise organization may require either one for all of its content
Account Deletion: Delete your account and all associated data (by default a 30-day grace period, then permanent removal; you can instead request immediate deletion)
Teamspace Deletion: Admin/owner choice — archive content to org or soft-delete; memberships, join requests, scoped tags/collections cleaned up
Organization Deletion: Cascading cleanup of all org data — files, tags, collections, members, teamspaces, integrations, media, RAG entries

7.3 Privacy Controls

Permission Management: Control who can access your content at org, teamspace, and file levels
AI Opt-Out: Disable AI processing per-file or globally; derived data deleted immediately
Integration Management: Connect or disconnect third-party services anytime; per-teamspace control
Notification Preferences: Per-channel (email, Slack, in-app) and per-category granular controls
Visibility Control: Set content visibility (private, organization, community, public)

8. Data Retention

8.1 Active Accounts

Data Type	Retention
User content (files, templates, agents)	Retained while account is active
Deleted files	By default a 30-day trash (recoverable), then permanent purge. Deleted immediately when you choose Delete Now, or when your organization requires immediate deletion. An Enterprise organization may instead require the 30-day trash only. A change to that setting applies to deletions made after it
Audit logs	90 days, then auto-deleted
AI operation logs (prompts, responses, token/cost/quality diagnostics)	400 days, then auto-deleted (retained for billing-audit and cost reconciliation); deleted immediately when your organization is deleted
Notifications	90 days, then auto-deleted
RAG retrieval analytics	90 days, then auto-deleted
Content-health metrics	90 days, then auto-deleted
Thread insights	180 days, then auto-deleted
Activity feed	180 days, then auto-deleted
IP addresses	Collected by infrastructure providers per their policies

8.2 Account Deletion

By default, account deletion enters the 30-day grace period below. You can instead request immediate deletion, which skips the grace period and removes your data right away. An organization's file deletion setting does not limit your right to delete your own account.

Phase	Timing	Action
Request	Immediate	Account marked for deletion, access revoked
Grace Period	30 days	Account recoverable; data preserved but inaccessible
Permanent Deletion	After 30 days	All data permanently removed — database records, file storage, RAG embeddings, audit entries, WorkOS identity

8.3 Deletion Workflow Detail

Account deletion follows a 13-step cascading process:

Pre-check: Block if user owns organizations with other members (ownership transfer required first)
Retrieve authentication provider ID (needed for auth cleanup)
Export user data (GDPR compliance — JSON snapshot before deletion)
Revoke all active sessions (force logout across all devices)
Delete authentication provider identity (prevents re-authentication)
Soft-delete then hard-delete all user files and storage blobs
Delete personal workspace/organization (cascading)
Remove all other organization memberships
Delete user media (avatars, uploads)
Delete AI agent threads and messages
Delete user tag associations
Create final audit log entry
Hard-delete user database record

Teamspace deletion offers two paths:

Archive to org: Files archived and unscoped to org level; tags/collections archived and unscoped
Soft-delete: Files enter the 30-day trash (or are deleted immediately where the organization requires it); tags/collections archived and unscoped
Both paths: memberships deactivated, join requests deleted

Organization deletion follows a comprehensive cascading process covering files, tags, collections, teamspaces (with member/join request cleanup), media, integrations, RAG entries, and all associated data.

9. International Data Transfers

9.1 Data Location

Component	Location
Frontend Hosting	Global edge network (Vercel)
Primary Storage	United States (Convex infrastructure)
CDN and Edge	Global distribution for performance (Vercel Edge Network)
AI Processing	United States (Google, Anthropic, OpenAI API endpoints)

9.2 Your Rights by Region

Region	Rights
GDPR (Europe)	Full data subject rights — access (Art. 15), rectification (Art. 16), erasure (Art. 17), portability (Art. 20), restriction (Art. 18), objection (Art. 21)
CCPA / CPRA (California)	Right to know, delete, correct, and opt-out of the sale or sharing of personal information. We do not sell your data. We use Google Analytics for product-usage analytics; you can opt out of that collection (see §11 — Cookies and Tracking)
Other Jurisdictions	We respect applicable local privacy laws

9.3 GDPR Compliance Implementation

Right to Access (Art. 15): exportUserData query returns all user data as JSON
Right to Erasure (Art. 17): requestAccountDeletion with a 30-day grace period by default (or immediate deletion on request), then complete removal
Right to Portability (Art. 20): Files downloadable in original formats; JSON data export
Data Processing: Enterprise data-processing arrangements are handled through our enterprise procurement process

10. Children's Privacy

Loop is not intended for users under 16 years of age. We do not knowingly collect information from children under 16. If you believe we have collected information from a child under 16, please contact us immediately at privacy@loopllm.ai.

11. Cookies and Tracking

Authentication Cookies: Session cookies for maintaining login state (required for service operation)
Analytics Cookies: We use Google Analytics to understand product usage and improve the service. Google Analytics is configured for analytics only — IP addresses are anonymized, Google Signals and advertising features are disabled, and analytics data is not shared with Google for advertising. You can opt out of Google Analytics using Google's Opt-out Browser Add-on.
Backend Telemetry: We use Axiom for backend performance and infrastructure telemetry (usage patterns, performance metrics, error traces) to operate and improve the service. Axiom telemetry does not use cookies and does not capture your content.
No Advertising Cookies: We do not use advertising cookies, retargeting pixels, or sell data to ad networks.

12. Data Breach Notification

In the event of a data breach affecting your personal information:

We will notify affected users without undue delay after confirming a breach, as required by applicable law
Notification will include: nature of breach, data affected, remediation steps, and contact information
We will notify relevant supervisory authorities as required by law
Contact for security incidents: security@loopllm.ai

13. Changes to This Policy

We may update this Privacy Policy to reflect changes in our practices or legal requirements. We will notify you of material changes via:

Email notification to your registered address
In-app notification
Update notice on our website

14. Contact Us

Privacy Questions

For privacy-related questions or data subject requests:

Privacy: privacy@loopllm.ai
Security: security@loopllm.ai
Legal: legal@loopllm.ai
Website: launch-engine.com

15. Legal Entity Information

Field	Value
Legal Name	Launch Engine, Inc.
Product	Loop (collaborative intelligence platform)
Jurisdiction	Delaware, United States
Website	launch-engine.com

16. Commitment to Privacy

At Loop, privacy is a core product principle, not an afterthought bolted on for compliance:

Privacy by Design: Built into every feature with granular per-file AI controls
User Control: You decide how your data is used, processed, and shared
Transparency: Clear about what we collect, why, and who processes it
Security First: AES-256-GCM encryption, WorkOS enterprise auth, SOC 2-certified infrastructure (Convex, Vercel)
No Data Sales: Your data is yours. We never sell, rent, or trade it.

Version History:

v2.10 — November 9, 2026: Added TypeSafe (AI decisions) as an AI provider and subprocessor (§4.3, §5.2); named Vercel's AI Gateway, set to zero data retention, as a route for AI requests (§4.3, §5.2); added transcription to OpenAI's rows and customer-sent email to Resend's (§4.3, §5.2)
v2.9 — October 29, 2026: Corrected the deletion terms (§7.2, §8.1, §8.2, §8.3, §9.3): the 30-day period is the default, not a guarantee. Files and accounts can be deleted immediately by the person's choice, and an Enterprise organization can require the 30-day trash only or immediate deletion only for its members' content (the setting never limits a person's right to delete their own account). Matches Terms of Service v2.8
v2.8 — October 28, 2026: Added Text Messages (SMS) (§3.4): account texts only, with opt-in, STOP and HELP, and no sharing of mobile numbers or text-message consent with third parties or affiliates for their marketing. Matches Terms of Service v2.7
v2.7 — October 26, 2026: Added Google User Data (§5.4): the Limited Use disclosure Google requires and the commitments behind it (feature-only use, no sale or ad use, no AI training, no human reading outside security, the law or your request, limited transfers). Excluded content that came from Google APIs from the beta review (§4.2). Named Gmail and Calendar, and the optional inbox feature, in the Google integration data (§2.3)
v2.6 — October 12, 2026: Removed the named default model from the AI provider table (§4.3). Added a sentence saying Loop configures standard models using Claude, Gemini, and OpenAI and that customers can change their default models or add their own API keys in organization settings (§4.3)
v2.5 — July 23, 2026: Accuracy pass against code. Re-disclosed Google Analytics as an active analytics subprocessor (§5.2, §11) and reversed the prior "GA replaced with Axiom" statement — GA is configured analytics-only (IP-anonymized, ad features off, no data-sharing), with a Google opt-out link; separated Axiom as cookieless backend telemetry (§11). Added LiveKit as a voice subprocessor (§5.2). Added a 30-day subprocessor-change notice (§5.2). Disclosed the 400-day AI-operation-log retention plus five previously omitted windows — notifications, RAG analytics, content-health (90d), thread insights, activity feed (180d) (§8.1). Removed the nonexistent "API request logs — 30 days" entry (§8.1). Replaced the unverifiable "150+ audit event types" claim with qualitative language (§2.4, §6.1). Corrected "TLS 1.3" to "TLS 1.2+" (Quick Summary, §6.1). Added teamspace to the visibility levels and CPRA sale/sharing opt-out to CCPA rights (§6.2, §9.2). Raised the minimum age to 16 (§10)
v2.4 — July 2, 2026: Clarified AI provider data handling (providers process under standard commercial API terms; no training on your content; limited safety/abuse retention per provider policy — replaces the absolute "no provider-side retention" language), reframed SOC 2 as vendor-certified infrastructure (Convex, Vercel) with Loop's own SOC 2 Type II in progress (§6.1, §16), genericized organization-deletion step count (§8.3); removed unverified commitments — fixed 72-hour breach-notification window softened to "without undue delay, as required by law" (§12), named Google Analytics replaced with accurate Axiom telemetry disclosure (§11; Axiom already listed as subprocessor in §5.2), DPA "available upon request" softened to enterprise-procurement handling (§9.3, §14)
v2.3 — May 8, 2026: Added beta review disclosure (§4.2), web search data transmission disclosure (§4.2), agent memory disclosure (§2.2), Axiom as subprocessor (§5.2), credit transaction logging (§2.2), observability/telemetry collection (§2.4), onboarding website scraping disclosure (§2.4), expanded HubSpot sync scope (§5.2), expanded AI operation log fields (§2.2, §4.2)
v2.2 — April 16, 2026: Added HubSpot/Twilio/Gmail API as subprocessors, Google Analytics disclosure, corrected deletion workflow steps (13-step), corrected audit event count, corrected IP address collection claim, corrected encryption scope
v2.1 — April 1, 2026: AI operation logging disclosure, provider data handling links updated, email standardization, HIPAA claim removed
v2.0 — March 10, 2026: Complete rewrite for WorkOS auth, actual subprocessor list, GDPR implementation details, teamspace scoping, deletion workflow documentation, data breach notification
v1.1 — September 18, 2025: Initial public privacy policy