FAQ

“Is my data safe?”: Yes. On a solo install it is stored locally on your device. On a team server it is stored on the server you host. Either way it is your machine, and we never see it.
“How do I back up my data?”: On a solo install: Download Backup writes a single encrypted file, automatic backups keep a copy in a folder you choose, and Vrge Cloud Backup adds off-machine copies we cannot read. On a team server your records live in the server's database, so backing it up is part of running the server — the self-hosting guide has the commands. Hosted backup for team servers comes after launch.
“Can I use Vrge on multiple devices?”: Solo: up to 2 devices, one active at a time. Team: one license activates the server you run, and people join by email invite up to your tier's seat count: 5 on Team Starter, 15 on Team Studio, 40 on Team Agency. The server turns down an invite past the cap, so pick the tier that fits and upgrade if you outgrow it.
“Does Vrge remind me to follow up?”: Better — it shows you what's waiting without you tracking anything. The Open Loops page lists every client thread waiting on your reply, every thread that went quiet after you wrote, and every overdue invoice — computed from who spoke last, not from reminders you set. When you answer the email or the payment lands, the loop closes itself. See the Open Loops guide.
“Does Vrge train on my data?”: No. Never. We do not ship your email or file content to any cloud AI for training. Inference-only calls for proposal refinement happen only when you've consented to a cloud provider. With your own key, Vrge sends the full text unless you turn redaction on. It is your provider account and your data, and Settings → AI shows you exactly what gets sent before it goes. Through Vrge Managed AI, content is always redacted and raw text can never be sent.
“Does Vrge work with Google Workspace (custom domain)?”: Yes. The full AI feature set works identically on Workspace and personal Gmail. Each team member connects their own mailbox through Settings → Sources → Gmail and signs in with Google, granting read-only access to mail. Vrge requests the gmail.readonly scope and cannot send, delete or modify anything. Workspace admins who prefer to keep the OAuth perimeter inside their own Google Cloud project can supply their own OAuth client in the connect dialog's Advanced section. The Connecting Sources page walks through the full flow including a note to forward to your IT admin if needed.
“Do I need internet to use Vrge?”: No. Core CRUD works fully offline. Internet is only needed for license activation, update checks, cloud AI calls (if you chose a cloud provider), and connected sources. Use Ollama locally for fully airgapped AI.
“What if the AI proposes something wrong?”: Reject it. Rejections genuinely teach it: Vrge tracks how often you accept each kind of proposal and quietly stops showing the kinds you keep turning down. If an accept turned out wrong, use the 24-hour undo. If a proposal is clearly a hallucination, hit “Report bad proposal” and pick a reason. That is recorded with the proposal so a wrong suggestion can be traced back to what produced it.
“Can I get a refund?”: License purchases are final for digital products. We recommend testing the app locally with Ollama (free) before purchasing. You can see the full proposal flow without any cost. Managed AI subscriptions can be cancelled any time through the Lemon Squeezy customer portal. Access continues through the end of your current billing period, then stops. No auto-renewal, no retention nags.
“Do I have to subscribe to Managed AI?”: No. It's entirely optional. Every feature in Vrge works identically when you bring your own OpenAI / Anthropic / Google key, or run Ollama locally, both free forever. Managed AI just means we handle the keys and billing for you. Not a single feature requires a subscription.
“What happens when I hit my monthly token cap?”: Your app falls back to BYO keys or Ollama for the rest of the month. No more managed calls until the next billing cycle. You will never see an overage bill. This is a hard quota, not “reasonable usage” weasel language. The proxy enforces it before calling any upstream provider.
“Does the proxy log my prompts or responses?”: No. The Managed AI proxy records only metadata: token counts, model name, task type, timestamp, redaction status. Prompt and response bodies are never persisted server-side. The privacy invariant is enforced at the schema level: there's no column for content in the usage log, so no code path can leak it. The proxy source is auditable.
“Which Managed AI plan should my team pick?”: This is about the optional Managed AI subscription, not the one-time Vrge licence. You only need it if you would rather not run a local model or bring your own AI key. Rough guidance: Starter ($10/mo, 1.5M tokens) covers about 15 new emails a day; Pro ($25/mo, 4M tokens) about 40 a day; Power ($75/mo, 12M tokens) about 120 a day. Email is the unit because only your mailboxes are read by the AI — bank, calendar and file events cost nothing. You can upgrade mid-cycle with a prorated quota credit, so start smaller and bump up if the admin dashboard shows you trending over cap.
“How do I stop one team member from using all our tokens?”: In team mode, an admin on any active tier sees a top-consumers table for the current period with a hard-cap box next to each member. By default everyone draws freely from the org quota. There is no automatic per-person split. Set a hard cap on a heavy user and the proxy stops them before the org quota is spent, so they can't starve the rest of the team.
“Can I self-host the Managed AI proxy?”: You do not need Managed AI at all. Run a local model on your own machine, or plug in your own cloud provider key, and no request is routed through ai.getvrge.com. Managed AI is the convenience path for people who would rather not set either up.