Frequently Asked Questions

In auto-reply mode, every response has a confidence score. If it falls below our threshold, the email is escalated to you or your team instead of being sent. In draft mode, the threshold is lower so you can review more suggestions before sending. You stay in control.

No. We only access the threads and messages needed to generate a reply. We don't read, store, or share your personal or sensitive data. Only encrypted response data is stored for your account, and we don't use your data for anything outside of serving your email automation.

No human reads your emails. Data is encrypted in transit and at rest. We follow strict security practices and don't store or share your email contents. Only you have access to your inbox and the responses InboxPilot generates.

All data from integrations is stored in isolated containers and encrypted at rest (AES-256) and in transit. We don't use your data for AI training or any purpose other than powering your InboxPilot account and supporting you when you ask.

An AI agent is a system that understands natural language and answers based on the information it's given. InboxPilot lets you build an AI that knows your business: you upload documents, FAQs, and content, and it learns from that to reply to emails in your voice and with your facts.

ChatGPT can't access your company's private data, so it can't answer accurately about your products or processes. InboxPilot combines similar AI capability with your own knowledge base—documents, FAQs, templates, and optionally a scan of your sent emails—so replies are accurate and on-brand.

Most conversations can be handled by the AI. When a question is unclear, sensitive, or beyond what the AI can answer confidently, the system can escalate so the customer can "talk to a person." The handover happens inside your existing tool (e.g. Gmail or Zendesk), and your team takes over from there.

No. Your data is never used to train external or public AI models.

The AI keeps replying while the conversation is relevant and within its capabilities. It may stop if the thread naturally concludes, if the request needs a human, or if you reply yourself—so you don't get duplicate responses.

Pricing is based only on the plan you choose. Each plan includes a monthly email allowance that resets on the first of each month. There are no hidden fees for normal use within your plan.

When a new email arrives, our system uses the content of the thread and the knowledge you've added (FAQs, templates, website, documents, and optionally what we learned from scanning your sent emails). The AI generates a reply that matches your tone and uses your business information. You can use draft mode to review before sending or auto-send for full automation.

Yes. The system can process and reply in multiple languages. If an email comes in another language, the reply is generated in that same language. For very specialized language needs, a human can step in.

Yes. InboxPilot uses the full thread context—previous messages and your knowledge base—to keep the conversation coherent and avoid repeating or contradicting earlier answers.

We escalate when we detect sensitive or complex issues, unclear requests, negative sentiment, or topics outside the AI's knowledge. We also escalate if the confidence score is too low, if the customer asks to speak to someone, or if we can't find a good answer. A team member can then review and respond.

Connect the email account that receives your form submissions (e.g. Gmail or Outlook) via Connect account in InboxPilot. In the setup, under optional reply-only keywords, add words or phrases that appear in those emails (e.g. "Contact Form" or "Support Request"). InboxPilot will then recognize form submissions and respond using your Knowledge Center and settings. Manage the inbox later in Reply settings & labels.

InboxPilot can scan your connected inbox (and sent folder) to learn your reply tone, style, and business context. From that scan we derive things like suggested templates and FAQs only when we see clear patterns—we don't invent data. This reduces how much you need to upload manually. You can run a scan from the Knowledge Center by selecting an inbox and using "Learn from my email."

When an inbox is first used or when you trigger "Learn from my email," we build that inbox's knowledge in the background. This can take up to a few minutes (sometimes up to 5). You can leave the page—when you come back to the Knowledge Center and select that inbox, the tabs will show the data once it's ready. We'll show a "Scanning this inbox" message until the build is complete.