OmniMendEffective: August 20, 2026
Version 2.3What changed in 2.3: We added Basic to the plans that use the existing hardware-ID license check. This does not add a new data category, purpose, recipient, or retention period.
Plain-English summary
- OmniMend is built around user-initiated diagnostics: we use the account, billing, device, and conversation data needed to run the repair service you ask for.
- Signed-in chat sync stores account-scoped copies of your chat history so your own chats can follow your account. It is separate from AI-improvement retention.
- Optional diagnostic improvement retention is gated by region, consent/default settings, sampling, and a global kill switch. When used, it stores redacted repair traces, not raw dumps or personal files.
- We never retain personal files, passwords, browser history from your device, keystrokes, raw memory dumps, SSH keys, or anything matching our secret-scanning filters.
- Security Vitals filters activity locally. Only a higher-risk event can reach cloud AI, and it uses a bounded, redacted event summary instead of raw file contents, full paths, or command lines.
- You can opt out or ask privacy questions by emailing privacy@omnimend.com. In-app privacy controls are planned for a future release.
- GDPR, UK-GDPR, CCPA/CPRA, and PIPEDA rights are honored.
Psynect Corp ("Psynect," "we," "us," or "our") publishes this Privacy Policy to explain how we collect, use, retain, and safeguard your information when you use OmniMend and related services (the "Service"). Please read it carefully. If you do not agree, you must not use the Service.
1. Information We Collect
1.1 Account Information
- Authentication data: email address and name (via Google OAuth or email signup)
- Account identifiers: unique user ID generated by our authentication provider (Supabase)
1.2 Device Information (from devices you choose to diagnose)
| Data Type | Purpose |
|---|---|
| Hardware ID (HWID) | License enforcement (Basic / Starter / Personal tiers) |
| System specifications & OS version | Diagnostic context for AI analysis |
| Crash logs, event logs, tool outputs | Root-cause analysis of issues you ask us to investigate |
| Driver information | Driver update recommendations |
| Device identifiers (e.g., IMEI, Apple serial) | Model-specific diagnostics you supply |
1.3 Conversation Data
- Messages you exchange with the AI agent
- Tool calls the AI makes and the tool results those calls return
- Session metadata (device type, session ID, mode)
- For signed-in users, account-scoped chat sync copies of conversation history and metadata so your own chats can sync across devices
1.4 Usage and Billing Data
- Internal usage transactions: internal credit changes, timestamp, model, input/output token counts, and cost
- Session heartbeats: for concurrent-session enforcement (auto-expire in 5 minutes)
- Payment metadata: processed by Lemon Squeezy; we receive order IDs and amounts but never your card number
1.4A Family Household Data
If you purchase, join, or are invited to an OmniMend Family, we process the household identifier, organizer and member account IDs, invited email addresses, invitation status and expiry, member role and status, registered computer identifiers and display names, subscription status, and shared usage and concurrency activity. We use this information to deliver the household plan, enforce its limits, prevent abuse, and provide the private Family management page.
The organizer can see active member email addresses, pending invitations, registered computer display names, and household-level plan usage. The organizer cannot use Family management tools to read another member's chat content, reports, diagnostic history, saved conversations, or Google account. Those records remain scoped to the member who created them.
1.5 Security Vitals Data
Security Vitals Beta runs on supported Windows PCs and works alongside the antivirus already protecting the PC. Monitoring, trust checks, and initial risk scoring happen locally. Most activity never reaches our cloud service.
When local scoring identifies a higher-risk event for cloud review, OmniMend sends a bounded event summary. Depending on the available evidence, that summary can include:
- an opaque event ID, a one-way device hash, the observation time, and the event source;
- the local risk band and code-based reasons that triggered review;
- coarse categories such as process family, parent-process family, path category, signature status, persistence type, and whether a relationship was locally trusted;
- bounded counts and yes/no security signals, such as public-connection count, known-antivirus detection, real-time-protection status, and whether network-delivery behavior was observed;
- for a second and final review round, a closed list of evidence codes returned by supported read-only probes.
The cloud event summary does not include file contents, full process paths, raw command lines, raw usernames, email addresses, or an unrestricted activity log. The cloud model returns a constrained assessment; OmniMend creates the customer-facing explanation from validated result codes.
1.6 Technical Data
- IP address and approximate country (from Cloudflare
cf-ipcountry) - User-Agent and app version
- Global Privacy Control (
Sec-GPC) header, if your browser or system sends one - Public website measurement data, such as page views, download clicks, checkout conversion events, referral codes, and basic campaign/source information
1.7 Public Website Analytics and Cookies
On public OmniMend website pages, we use Google site measurement and conversion tags, plus first-party referral cookies, to understand which pages, downloads, referrals, and checkout flows are working. If a checkout or sign-in redirect includes an email address, the page may pass that address to Google tag's conversion attribution tools. This website measurement is separate from the OmniMend desktop diagnostic workflow and is not used to read device files, app diagnostics, or chat contents.
1.8 What We Do NOT Collect
- Personal files, documents, photos, or media on your devices
- Passwords or credentials stored on your system
- Browser history from your device
- Keystroke data or screen recordings
- GPS location data
- Contents of memory dumps — we extract structured diagnostic fields only; raw binaries never leave your machine
2. How We Use Your Information
| Purpose | Data Involved | Legal Basis (GDPR / UK-GDPR) |
|---|---|---|
| Deliver AI diagnostics and repairs | Account, conversation, device data | Contract (Art. 6(1)(b)) |
| Process payments, manage subscriptions | Account, billing | Contract (Art. 6(1)(b)) |
| Enforce licenses & concurrent-session limits | HWID, session metadata | Contract + legitimate interest |
| Prevent fraud, abuse, security incidents | Technical, billing, account | Legitimate interest (Art. 6(1)(f)) |
| Deliver Security Vitals cloud review | Account entitlement plus the bounded, redacted security event fields described in §1.5 | Contract (Art. 6(1)(b)); legitimate interest in protecting users and the Service (Art. 6(1)(f)) |
| Improve OmniMend diagnostics and repair models | Redacted conversation + tool traces — only when permitted by §4 | Consent or opt-out default where legally permitted (Art. 6(1)(a) / applicable local law) |
| Support and service communications | Account, conversation context | Legitimate interest |
| Measure public website and checkout performance | Website events, referral/campaign data, checkout attribution, browser/device metadata | Consent where required; legitimate interest or business-purpose processing where permitted |
| Comply with legal obligations | As required | Legal obligation (Art. 6(1)(c)) |
3. Data We NEVER Retain
Regardless of your consent settings, the following categories are never stored on our servers. A deterministic scrubber runs at the edge before any retention write:
- Raw
.dmpminidump binaries (we extract structured fields only) - Shell command output (arbitrary — may contain private files)
- File contents retrieved by arbitrary-path read commands (Mac
read_file, etc.) - Strings matching regex patterns for SSH private keys, PEM blocks, JWTs, API keys, bearer tokens,
password=,secret= - Full Android logcat buffers (notification text, OAuth URLs)
- Lists of installed applications when the request originates from the EU, UK, EEA, Switzerland, Brazil, India, or South Korea (because app names can reveal special-category data like religion, health, or orientation)
- Free-text email addresses inside tool output (scrubbed to
<email>placeholder) - Hostnames and user paths are replaced with pseudonymous tokens before retention
- Security Vitals raw event streams, full process paths, raw command lines, and file contents
4. Optional Diagnostic Improvement Retention
4.1 What optional diagnostic improvement means
When optional diagnostic improvement retention applies, we keep a redacted copy of selected diagnostic conversations and tool traces. This is used only to improve OmniMend's repair quality, not for advertising or unrelated profiling. We use these traces to:
- Improve the OmniMend AI's ability to diagnose and fix problems
- Train or evaluate repair-focused models on patterns of what works and what does not
- Build a library of crash-to-fix signatures that makes the product faster and more accurate for everyone
- Create aggregated research about device failure trends, without identifying individual users
4.2 How your consent default is determined
OmniMend applies different defaults based on where your request originates (determined by Cloudflare's cf-ipcountry header). If the global retention kill switch is off, if sampling is set to zero, or if the required redaction/storage pipeline is unavailable, no optional diagnostic improvement copy is stored regardless of the table below.
| Region | Default | How to change |
|---|---|---|
| United States, Australia, New Zealand, Japan, Singapore, Hong Kong, Peru, and similar jurisdictions after legal review | Opt-out default where active. Optional diagnostic improvement may be enabled by default where legally permitted. GPC (Sec-GPC: 1) is treated as automatic opt-out. |
Email privacy@omnimend.com or POST https://crashpro-inference.psynect.workers.dev/v1/privacy/dsr. In-app privacy controls are planned for a future release. |
| United Kingdom, European Union, European Economic Area, Switzerland, Brazil, India, South Korea, Canada, Mexico, China, Thailand, Indonesia, Vietnam, Nigeria, Turkey, Russia, Saudi Arabia, Colombia, Argentina, United Arab Emirates, South Africa, Israel, Taiwan, Chile, and other strict opt-in jurisdictions | Opt-in required. Data contribution is OFF. Nothing retained until you affirmatively opt in. | To opt in: email privacy@omnimend.com. In-app privacy controls are planned for a future release. |
| All other regions | Opt-in default. Safest treatment when the region is not clearly classified. | To opt in: email privacy@omnimend.com. In-app privacy controls are planned for a future release. |
4.3 Permitted diagnostic-improvement purposes
Optional retained traces may be used only for these repair-focused purposes:
- Diagnostic retention for repair-model improvement — retain redacted diagnostic traces to train or evaluate device-repair diagnostics. Training uses redacted traces only; no raw PII reaches the training pipeline.
- Benchmarking — evaluate diagnostic accuracy and tool success rates across the user base (aggregated statistics only)
- Crash pattern analysis — correlate BSOD/crash signatures with successful fix actions to improve tool design
We do not use optional retained diagnostic traces for advertising, behavioral profiling, data brokerage, or unrelated product lines.
4.4 Voluntary participation
Declining optional diagnostic improvement retention does not block core repair features. If we offer future features that rely on the historical repair corpus, we will describe that dependency clearly before you choose to use them.
4.5 Withdrawing consent
You may withdraw any consent at any time without affecting access to core features. Methods:
- Email privacy@omnimend.com (we process withdrawal within 72 hours; internal SLA)
- Submit a Data Subject Request via
POST https://crashpro-inference.psynect.workers.dev/v1/privacy/dsr(see §8) - In-app privacy controls are planned for a future release. Until then, email is the fastest path.
After withdrawal:
- New sessions will not be retained from the moment of withdrawal
- Previously retained data is deleted within 30 days (cryptographic shredding within 72 hours; aggregate-index cleanup within 30 days)
- Aggregated statistics already derived from your data cannot be retroactively unaggregated, but no further individual processing takes place
5. Third Parties Who Process Your Data
5.1 AI Model Providers
- Google (Gemini): primary AI provider — Google AI Terms
- Anthropic (Claude): fallback provider — Anthropic Privacy Policy
- OpenAI (GPT): tertiary fallback — OpenAI Privacy Policy
- DeepInfra: configured online provider for Security Vitals cloud review — DeepInfra Data Privacy
These providers process data under their own privacy policies. Where zero-retention API modes are available, we use them. We do not instruct these providers to retain your data for our purposes.
5.2 Infrastructure Providers
- Supabase: authentication, user database, consent ledger, billing records
- Cloudflare: inference relay (Workers), retention infrastructure (D1, KV, R2, Vectorize), CDN
- Lemon Squeezy: payment processing (PCI DSS compliant)
- Google site measurement: public website analytics, conversion attribution, and advertising measurement
5.3 PII-Redaction Providers (Retention Pipeline Only)
If you have consented to diagnostic retention, each conversation is sent through an LLM-based redaction service before it is stored. The redaction service replaces personally identifiable information (emails, usernames, hostnames, device identifiers, IP and MAC addresses, phone numbers, account names, file paths containing usernames, and similar) with typed placeholders such as [REDACTED-EMAIL] or [REDACTED-UDID].
To keep this process reliable, we route the redaction request through a configured fallback chain and fail over automatically if one provider is unavailable. The exact order may change as providers, data-processing terms, and reliability change. Current or optional redaction providers may include:
- Google (Gemini): redaction and diagnostic AI provider — Google AI Terms
- OpenAI (GPT): redaction and diagnostic AI provider — OpenAI Privacy Policy
- Anthropic (Claude): redaction and diagnostic AI provider — Anthropic Privacy Policy
- Groq: optional redaction fallback if configured and legally appropriate — Groq Terms of Use
Data sent to a redaction provider is used only to return a redacted version of the content. Where zero-retention or no-training API terms are available, we use them and do not instruct providers to retain content for their own purposes.
Fail-closed guarantee. If every redaction provider in the chain is unavailable and all automatic retries have been exhausted, the conversation is not stored. We would rather lose a single data point than keep any content that has not successfully passed through the redaction pipeline. An automated alert notifies our team within seconds of any chain-wide failure so we can investigate immediately.
All data in transit uses TLS 1.3. Data at rest in Cloudflare and Supabase is encrypted with AES-256. Content blobs containing retained diagnostic traces receive an additional per-user encryption layer using HKDF-derived keys and AES-GCM, so even a hypothetical bucket-level breach would not expose plaintext content.
5.4 Things We Do NOT Do
- Sell your personal information
- Share your diagnostic data with advertisers
- Use your data for targeted advertising
- Provide your data to data brokers
6. Data Retention Periods
| Data Type | Retention |
|---|---|
| Account data (email, user ID) | While your account is active; deleted within 30 days of deletion request |
| Billing / transaction history | 3 years (accounting and audit obligations) |
| Session heartbeats | Auto-expire after 5 minutes of inactivity |
| Family membership, invitation, and computer-registry data | While the Family subscription or membership is active, plus limited audit and billing retention as reasonably required for security, dispute resolution, accounting, and legal obligations. Expired invitation secrets are not displayed and are replaced when an invitation is resent. |
| Consent log entries | Life of account + 3 years (proof-of-consent) |
| Account chat-sync cloud copies (signed-in users) | While your account is active and account sync remains enabled in the current product; deleted within 30 days of verified account erasure or DSR. Local-only deletion may hide the chat on that device before every cloud copy is fully removed. |
| Public website analytics and conversion records | Handled by our website measurement providers according to their retention settings and applicable law; internal attribution records are kept only as long as reasonably needed for analytics, accounting, fraud prevention, and marketing measurement. |
| Retained diagnostic traces (if consented) | Up to 24 months from creation, or sooner on withdrawal |
| Security Vitals bounded investigation records and constrained results | Targeted for deletion after 30 days, with a 31-day hard boundary for replay-safe cleanup; verified erasure requests remove account-linked security records sooner |
| Security Vitals daily request, token, and cost totals | Up to 35 days for abuse controls, cost reconciliation, reliability review, and aggregate service reporting |
| Aggregated research statistics derived from consented traces | Indefinite, provided the aggregation satisfies the EDPB three-part anonymisation test (no singling out, no linkability, no inference). Underlying pseudonymous records remain subject to the 24-month cap. |
| DSR request records | 3 years (regulatory compliance) |
7. Security
- All data in transit: TLS 1.3
- All data at rest: AES-256 (Cloudflare + Supabase)
- Retained content blobs: additional AES-GCM layer with per-user HKDF-derived keys
- Authentication: JWT with HMAC-SHA256 verification; secrets in Cloudflare Workers encrypted store
- Rate limiting, anomaly detection, and abuse prevention on Worker ingress
- Electron renderer has no direct Node.js access (context isolation)
- Security disclosure contact: support@omnimend.com; bug bounty scope includes privacy issues
7.1 Breach Notification
If we discover a personal data breach, we will notify the relevant supervisory authority within 72 hours of becoming aware of it (where required by GDPR Art. 33 / UK-GDPR). We will notify affected users without undue delay where the breach is likely to pose a high risk to their rights and freedoms (GDPR Art. 34). Our notice will describe the nature and scope of the breach, the categories of data affected, and the measures we have taken.
8. Your Rights
| Right | What it does | Applicable regions |
|---|---|---|
| Access | Copy of data we hold about you | All |
| Rectification | Correction of inaccurate data | All |
| Erasure | Delete your personal data | EEA, UK, CA (CCPA) |
| Portability | Machine-readable export | EEA, UK |
| Objection | Object to certain processing | EEA, UK |
| Restriction | Pause processing during dispute | EEA, UK |
| Withdraw consent | Stop consent-based processing going forward | All |
| Opt-out of sale / sharing | CCPA / CPRA opt-out | California |
| Non-discrimination | We will not retaliate for exercising any right | All |
8.1 How to Submit a Data Subject Request
- Email: privacy@omnimend.com with your registered email and the request type (access, erasure, portability, etc.)
- API:
POST https://crashpro-inference.psynect.workers.dev/v1/privacy/dsr(authenticated with your account token) - An in-app Settings → Privacy → "Request My Data" button is planned for a future release. Until then, email is the recommended path.
For erasure requests after you have deleted your account, we accept email-link verification so you can still exercise the right without an active login.
Requests are fulfilled within 30 days. Our internal SLA targets 72 hours for most cases.
9. Children's Privacy
OmniMend is intended for users 18 years of age or older. Each user must affirm age at signup via a checkbox in our setup wizard. We do not knowingly collect or retain personal data from users under 18. If we discover, or are notified by any party, that a user under 18 has created an account, we will delete all associated data within 7 days and close the account. Notify privacy@omnimend.com if you believe a minor has created an account. Where local law sets a higher digital-consent age (for example 16 in some EU member states), that age applies in that jurisdiction instead.
10. International Data Transfers
Your data may be processed in the United States and other countries where our infrastructure providers operate. For transfers out of the EEA, UK, or Switzerland, we rely on Standard Contractual Clauses (EU Commission 2021), the UK International Data Transfer Addendum, and processor certifications (SOC 2 Type II, ISO 27001) where available. By using the Service, you acknowledge these transfers.
11. Region-Specific Disclosures
11.1 California (CCPA / CPRA)
Under the California Consumer Privacy Act and CPRA, you have the right to know, delete, correct, limit use of sensitive personal information, opt out of sale/sharing, and be free from retaliation. We do not sell personal information. We disclose diagnostic conversation data to configured AI model providers (including Google, Anthropic, or OpenAI) and bounded Security Vitals event summaries to DeepInfra solely to deliver the AI inference service you requested. We also use public website measurement providers for analytics and conversion attribution. These disclosures are made to service providers under contractual data-use limitations and are not intended as a "sale" or "sharing" for cross-context behavioral advertising under CPRA §1798.140. You may still exercise opt-out and sensitive-personal-information rights by emailing privacy@omnimend.com or using the "Your Privacy Choices" link in our website footer. We honor the Global Privacy Control (Sec-GPC) header as an opt-out signal for optional diagnostic improvement retention where applicable and display an "Opt-Out Signal Honored" indicator on this page where it is detected.
Categories collected in the past 12 months: identifiers (email, user ID), internet activity (Service usage), professional information (Team/Enterprise tiers), geolocation (country only), inferences (diagnostic patterns — only if you consented).
11.2 Europe / United Kingdom (GDPR / UK-GDPR)
Legal bases for processing are summarized in §2. You have the rights listed in §8.
- Data Controller: Psynect Corp, contact privacy@omnimend.com
- Data Protection Officer: OmniMend does not currently require a mandatory DPO under GDPR Art. 37. Privacy inquiries and data subject requests are handled by Psynect Corp at privacy@omnimend.com. We will appoint a qualified DPO when required by applicable law or operational scale.
- Supervisory authority: you may complain to your local authority (UK ICO at ico.org.uk, Irish DPC at dataprotection.ie)
11.3 Canada (PIPEDA)
We process personal information with your consent (express for retention, implied for service delivery). Contact privacy@omnimend.com for access, correction, or deletion. You may complain to the Office of the Privacy Commissioner of Canada.
11.4 Australia / New Zealand
We comply with the Australian Privacy Principles (Privacy Act 1988) and the NZ Privacy Act 2020. Requests to privacy@omnimend.com; complaints can be escalated to the OAIC (Australia) or the Office of the Privacy Commissioner (NZ).
12. Changes to This Policy
We may update this Privacy Policy. For material changes (new data categories, new processors, new purposes), we will:
- Notify you by email and in-app at least 14 days before the change takes effect
- Require fresh consent from users in opt-in regions for any new processing purposes
- Post the new version with a new "Effective date" and updated document hash
- Maintain prior versions or a change history when required by law or when a material change needs a user-facing archive
Continued use of the Service after a non-material change constitutes acceptance of the updated policy. For material changes in opt-in regions (UK, EU, etc.), no retention under new purposes occurs until you affirmatively consent.
13. Contact
Psynect Corp
Privacy: privacy@omnimend.com
Security: support@omnimend.com
Support: support@omnimend.com
Website: omnimend.com
Appendix A — Exact Fields Retained Under "Diagnostic Retention" Consent
When you have consented to diagnostic retention, we keep the following per session:
| Field | Description | PII Class |
|---|---|---|
user_id_hash | One-way hash of your Supabase user ID | Pseudonymous |
session_id | Opaque session identifier | Non-PII |
device_type | pc / mac / iphone / android / chromebook / nopower | Non-PII |
mode | diagnosis / repair | Non-PII |
country_bucket | coarse bucket: us / eu / other | Non-PII |
app_version | OmniMend version | Non-PII |
tool_sequence | ordered list of {tool_name, input_summary, output_summary} | Redacted; no free text, secrets scrubbed |
initial_complaint_category | classifier output: bsod / slowness / wifi / boot / etc. | Non-PII |
repair_applied_bool | whether any repair tool was called | Non-PII |
resolution_signal | implicit: was a new session opened for the same device within 72h | Non-PII |
redacted_transcript | conversation content after configured redaction has succeeded | Redacted |
The redacted_transcript blob is stored on Cloudflare R2 under an additional per-user encryption layer. Deleting the per-user key (on erasure requests within 72 hours) renders the blob unreadable.
Appendix B — Security Vitals Cloud Event Summary
Security Vitals cloud analysis uses a closed schema. The service rejects unknown top-level fields and does not accept free-form prompts from the desktop client. Current field groups are:
| Field group | Examples | Privacy treatment |
|---|---|---|
| Event identity | Opaque event ID, one-way tenant/device/session hashes, observation time, source | Pseudonymous; no raw account or device identifier in the model request |
| Local decision | Risk band, reason codes, novelty/impact routing bands, telemetry trust | Code-derived and bounded |
| Process context | Coarse path category, process family, parent family, signature status, relationship-trusted flag | No full path or raw command line |
| Behavior signals | First-seen/seen-days ranges, persistence types, bounded connection count, antivirus and real-time-protection flags | Categorical, boolean, or bounded numeric fields |
| Optional round-two evidence | Closed evidence codes from supported read-only probes | No arbitrary probe output or file content |
| Operational usage | Provider/model outcome, input/output tokens, estimated provider cost, latency | Used for reliability, limits, and cost monitoring |