Skip to content
Vrge

Last updated: July 30, 2026

Trust Center

This page is the single index of Vrge's security, privacy, and compliance posture. It is built for procurement, security, and legal teams who need to evaluate Vrge before purchase. We are honest about what is in place today, what is in progress, and what we do not hold. We do not display certifications we have not earned.

The short version

Vrge is local-first by design — customer business data lives on customer devices. Source connections (Gmail, Calendar, Drive, banking) are read-only. Cloud AI calls are redacted by default. AI proposals never write to the database without explicit user approval. Nothing is sold, no telemetry exfiltrates customer data, and one-click export works even with an expired license. The rest of this page is the receipts.

Architectural commitments

The structural decisions that make Vrge a different category of risk than a typical cloud CRM. These are properties of the codebase, not promises in a policy.

Local-first storageIn place

Customer data lives in a SQLite database on the user's own device. Vrge does not host customer business data on its own servers.

Read-only access to every sourceIn place

Read-only access to every connected source. Where a provider offers a read-only OAuth scope, Gmail, Calendar, Drive — Vrge requests only that. Where none exists (Stripe Connect, QuickBooks Online) Vrge requests the narrowest scope available and enforces read-only in the client, which issues GET requests against an allow-list and never writes. Vrge cannot send email, modify calendars, or alter source data.

You choose what any cloud AI seesIn place

Local Ollama sends nothing off the device. With the customer's own cloud key the default is full content — their provider account, their data — and redaction modes (tokenised, locally summarised, or truncated) can be set globally or per source, each with a 'what gets sent' preview. Vrge Managed AI is floored to redact-by-default and can never transmit raw content.

Proposal review before any writeIn place

AI-detected entities never write to the customer's database directly. Every observer-generated change requires explicit user approval through the Inbox.

24-hour reversible acceptIn place

Every accepted proposal can be undone within 24 hours via transactional rollback. The applied_proposals log records full snapshots for that purpose.

Every record shows where it came fromIn place

Each entity created via the observer carries an origin record: source event, AI model, confidence score, accept time. Auditable end-to-end.

One-click data portabilityIn place

Full export to portable JSON/CSV: entities, proposals, AI history, origin records. Works even with an expired license. No DRM, no lock-in.

No training on customer dataIn place

Customer email, file, or business data is never submitted to any AI provider for training. Inference-only API calls, when authorised by the user.

Security posture

Concrete technical and operational measures, including what isn't in place yet.

TLS in transitIn place

All traffic between Vrge endpoints (website, license, auto-updater, Managed AI proxy) and customer devices uses TLS 1.2 or higher.

Bcrypt password hashingIn place

On the optional Team self-hosted server, user passwords are bcrypt-hashed with per-user salts. JWT for session management with rotation.

OAuth tokens stored only on-deviceIn place

No Vrge-operated server retains a token; under a Team license they sit encrypted on your own server. For Gmail, Google Calendar and Outlook they never leave the device at all. Plaid, Stripe, Square and QuickBooks refuse to issue tokens to an app that cannot keep a secret, so a Vrge relay performs the exchange and refresh in transit — and for Plaid the token accompanies each read — storing neither.

Cookieless analyticsIn place

The marketing site uses Cloudflare Web Analytics — aggregate page views and coarse geography only, no cookies, no fingerprinting, no personal identifiers.

Coordinated vulnerability disclosureIn place

Triage commitment within one business day of report. Public scope and rules of engagement on the Security page. Safe-harbour language for good-faith research.

Security & disclosure policy

Independent security auditNot held

Vrge has not commissioned a formal third-party penetration test. We do not claim one, and we will not display a badge we have not earned.

Google CASA security assessment (Tier 2)In place

Required by Google for OAuth verification at restricted Gmail and Calendar scopes. Vrge completed a Tier 2 assessment with an authorised lab and passed with no security findings. Letter of Validation on file; available to enterprise buyers on request.

SOC 2 Type INot held

Vrge is not SOC 2 certified. We say so plainly rather than implying a posture we do not hold, and we will not display badges we have not earned.

Frameworks Vrge addresses, with links to the underlying documents your legal team will want to read.

GDPR (EU/UK/Swiss)In place

Article 28 controller–processor template available. Standard Contractual Clauses (Module 2) incorporated. Sub-processor change notification with 30-day objection window.

Data Processing Agreement (DPA)

CCPA / CPRA (California)In place

Privacy policy details CCPA categories of information collected. We do not sell or 'share' personal information as those terms are defined under CCPA.

Privacy Policy — California section

HIPAAIn place

TRD Ventures LLC is not a HIPAA-covered entity, and Vrge is not a HIPAA-compliant system today. Customers in healthcare should not store, process, or transmit Protected Health Information through the Service.

Privacy Policy — HIPAA notice

Sub-processor disclosureIn place

Complete current list of third-party services that may process data on Vrge's behalf, with purpose, data categories, and jurisdiction for each.

Sub-processors

PCI DSSIn place

Vrge does not collect, transmit, or store payment card data. All payments are processed by Lemon Squeezy as Merchant of Record — they are PCI DSS Level 1 compliant.

Data residencyIn place

Customer business data resides where the customer chooses — their device, their self-hosted server. Vrge-operated services run primarily in the United States; the marketing site is on Cloudflare's global network.

For security and procurement teams

We respond to security questionnaires (CAIQ, SIG, custom), DPA redline requests, and procurement intake forms within five business days. For most B2B engagements the published policies here are sufficient — counter-signed PDFs are available on request.

DPA / privacy
dpa@getvrge.com

Documents