What is the operational difference between Draft Mode and Auto-Send Mode?

Draft Mode functions as a manual quality gate by generating context-aware email replies and placing them directly into your outbox for human review. Auto-Send Mode provides autonomous execution by bypassing human sign-off to transmit background email responses instantly for routine, low-stakes communication categories.

How does Form-Only Mode process website landing page submissions?

Form-Only Mode allows you to standardize incoming omnichannel data by routing website forms through the AI engine. By applying the command 'Only handle website forms' in your specific inbox settings panel, the AI strictly isolates its operational processing to incoming web leads, ignoring general email chains.

Are team member inboxes segregated or visible to others within InboxPilot?

InboxPilot enforces absolute workspace segregation and data isolation. Every connected inbox functions as a strictly private silo with its own standalone Knowledge Center, private activity logs, and unique logic configurations. No team member can access, view, or audit another colleague's prompt setups or execution logs unless an environment is explicitly configured as a Shared Inbox.

Is my company data used to train public generative AI models?

No. InboxPilot utilizes strict data isolation framework protocols. Your emails, internal knowledge bases, configuration records, and customer interactions are securely siloed within your dedicated application environment and are never shared with public LLMs or utilized to train external public generative databases.

Help Center | Inboxpilot

InboxPilot is an AI-powered email automation platform engineered to manage customer support, high-volume operations, executive gatekeeping, and inbound sales pipelines. By processing incoming messages through an isolated, business-specific Knowledge Center, the platform automatically generates context-aware drafts, routes messages based on semantic intent, and executes background workflows directly within your existing email workspaces.

1. The Core Implementation Blueprint (3-Step Quick Start)

Step 1

Train the Knowledge

Ingest documentation, live system URLs, raw text FAQs, or background auto-scan your sent message history.

Step 2

Secure the Inbox

Establish sandboxed OAuth connections to Gmail, Outlook, or Zendesk instances with zero password sharing.

Step 3

Simulate & Deploy

Stress-test workflows safely in the Playground, track real-time logs, and deploy plain-English inbox rules live.

Step 1: Train Your Knowledge Center

To ensure your AI email agent generates factually accurate responses aligned with your distinct brand voice, populate the Knowledge Center (accessible via the left sidebar). The system utilizes six distinct tabs to ingest information:

AI Learned: Displays structural communication patterns, tone preferences, and context dynamically harvested from your inbox history.
Email Templates: Retains custom macros. Write a template once for repetitive questions, and the AI deploys that specific framing whenever it detects matching user intent.
FAQ: Houses structured question-and-answer pairs for direct troubleshooting.
Files: Ingests internal policy sheets, operational playbooks, and documentation (PDF, TXT, DOCX).
Website: Automatically crawls and indexes public product pages or external documentation URLs.
Raw Text: Accepts copy-pasted operational updates, temporary pricing tiers, or conditional guidelines.

💡 Automated Alternative: Turn on the background email auto-scan feature. The platform will automatically parse your connected inbox's sent folder to map your organic tone, writing style, and recurring responses with zero initial manual input.

Step 2: Connect Workspaces via OAuth

Navigate to the Connect Account panel to link your inbox environments. InboxPilot uses sandboxed OAuth integrations, ensuring your corporate system passwords are never processed, viewed, or stored.

Gmail & Google Workspace: Supports personal aliases and large corporate distribution groups.
Outlook & Microsoft 365: Connects individual enterprise seats or shared corporate mailboxes.
Zendesk: Syncs directly with customer service help desks to automate support ticket responses.

Step 3: Run Sandbox Simulations & Go Live

Before activating autonomous operations on real incoming customer messages, use the Playground tab to stress-test your logic.

Paste sample emails into the sandbox to verify reply accuracy, tone matching, and classification tags.
Tweak instructions and re-run simulations until the output matches your performance guidelines.
Flip your settings live and use the Logs tab to monitor background actions, confidence scores, and processing states in real-time.

2. Dynamic Operational Configurations

Customize how the AI operates by modifying these granular controls inside your Reply Settings & Labels panel:

Draft Mode	Manual Quality Gates	Generates context-aware responses directly inside your thread for human approval.
Auto-Send Mode	Autonomous Execution	Bypasses human sign-off to transmit background replies instantly for routine message categories.
Temperature Control	Style Regulation	Lower settings (0.0–0.3) force conservative, strictly factual outputs. Higher allows creative variations.
Semantic Triggers	Target Filtering	Directs the AI to process or respond only when specific conditions, tones, or intent parameters are identified.
Skip Keywords	Automation Safeguards	Instantly stops automated actions on any thread containing specific terms like "legal", "urgent", or "refund".
Form-Only Mode	Omnichannel Inbound	Type `Only handle website forms` into settings to restrict the AI explicitly to web landing page form submissions.

3. Workplace Versatility: Target Use Cases

InboxPilot adapts to distinct organizational roles and vertical business models through targeted, plain-English instructions.

Executive Offices & Assistants

Inbox Gatekeeping: Acts as an automated layer for founders and executives, filtering out cold sales pitch noise and highlighting high-priority investor or board threads.
Administrative Routing: Automatically identifies calendar booking requests, passes receipts to accounting, and flags schedule updates for human EAs.

Customer Support & Operations

24/7 Support Tier: Resolves repetitive product questions, common FAQs, and troubleshooting issues instantly.
Task Automation: Standardizes backend ticket triage, processes refund requests, manages order updates, and routes technical escalations based on severity.

Dedicated Sales Teams

Lead Optimization: Scores and processes incoming contact form leads, matches prospective buyers to pricing matrices, and sends calendar scheduling links.
Follow-Up Automation: Keeps prospects warm by drafting contextual follow-ups based on the sales context stored in your CRM (e.g., Pipedrive, Notion).

4. Privacy, Security & Data Isolation Controls

InboxPilot is built from the ground up to support enterprise-grade security and corporate compliance guidelines.

Strict Workspace Segregation: Every single connected inbox is fully isolated. Each team member has a private Knowledge Center, private data logs, and independent configurations. No user can access or view another team member's configuration panels unless an environment is explicitly marked as a Shared Inbox.
Zero Public Model Training: Your data belongs entirely to you. InboxPilot enforces a strict privacy silo; your emails, uploaded internal documents, customer interactions, and configuration preferences are never shared with public LLMs or used to train external models.
Banking-Grade Security: Data streams are securely isolated using industry-standard protocols, protected by encryption frameworks both at rest and in transit.