Policy-Enforced Responses
AI responses that never expose unauthorized data, governed by your company policies.
Complete guide to building, deploying, and managing your AI-powered agents
Get your first AI agent up and running in under 5 minutes. Follow our step-by-step guide.
Enterprise-grade conversational AI
6-Step Wizard: Info → Knowledge → Personality → Channels → Appearance → Deploy
Navigate to AI Employees → New AI Employee. The wizard walks you through 6 steps:
Your AI agent is live immediately after creation and starts responding to messages on all connected channels.
Documents, URLs, Sitemaps, APIs
There are three ways to add knowledge. All content is chunked, embedded, and retrieved with source citations at chat time.
Upload one or more files in a single drag-and-drop. Supported formats: .pdf, .docx, .txt, .csv, .md. Maximum 10 MB per file. Unsupported types (.xlsx, images, .zip, etc.) are blocked at the upload dialog with a clear error message so you never wait through a broken upload.
Scrape any public webpage or bulk-import from a sitemap. Click Add URL and paste the address, or Import Sitemap with a sitemap.xml URL to ingest an entire site at once.
Paste content directly via Add Text. Useful for FAQs, policies, or short answers that don't warrant a full document.
The agent is hard-coded to refuse enumeration attempts like "list all files", "how many documents", or "show me your first document". These prompts are caught before retrieval, so the AI never receives file names for meta questions — and never leaks them to visitors.
Control how your agent responds
Rules are instructions that control how your AI agent responds. Unlike the system prompt, rules are specific directives.
Rules at the top have higher priority when conflicts occur.
Invite up to 2 moderators per agent
Invite team members to help manage specific AI agents without giving them full account access.
Limit: 2 moderators per AI agent. Ownership can be transferred, but base account permissions stay with the account owner.
Session policy · Trusted Devices · Role permissions
Harmonix follows the same session model as GitHub, Linear, and Vercel:
Every device you sign in from is tracked under Settings → Security → Trusted Devices. From there you can:
You can't revoke your current device — use the standard Sign Out button for that.
The Active Sessions card lists every browser you're currently signed in on. Use Sign out of all other devices to keep your current session and drop every other one at once.
The first time an unrecognised device signs into your account you'll receive an email at your registered address showing the approximate location, browser, and time. Trusting the device silences future alerts for it.
System admins can invite sub-admins and grant fine-grained access via Admin → Users → Shield icon. Per-user permission overrides layer on top of the assigned role — so you can grant one moderator write access to Roles without promoting them to full System Admin.
Auto-syncs working hours, timezone & vacation from Google Calendar
Allow visitors to schedule appointments through your AI agent. The system reads your actual Google Calendar to determine your availability — no manual setup required.
Connect to external systems
Harmonix hooks into your stack in three ways: outbound webhooks (fire when events happen), inbound API calls to enrich agent responses, and native channel connectors (WhatsApp, Slack, etc.). Webhooks are configured per-agent under the Integrations section — the older "Add URL" knowledge tab has been retired.
Every webhook body is signed with HMAC-SHA256 in the X-Harmonix-Signature header so your server can verify authenticity. Retries are automatic with exponential back-off; failed deliveries surface in the Integrations tab.
Common uses: sync leads into your CRM (Salesforce, HubSpot, Pipedrive), trigger Zapier / n8n workflows, or post to internal Slack channels.
One line of code to embed
Use the Theme tab to customize colors, border radius, and position of your chat widget.
Colors, position & branding
Make the chat widget match your brand by customizing its appearance.
Preview your changes in real-time with the Test Chat button.
Refer others & earn
Every user can earn money by referring others to Harmonix through our Affiliate Program.
The more referrals you bring, the higher your commission rate:
Request a payout once your pending balance reaches $50. Payments are processed via Stripe.
You'll receive email notifications whenever you earn a commission or reach a new tier!
Floating Button or Side Tab — choose your icon & label
Go to your agent's Appearance tab to fully customize how the chat button looks on your website.
Pick from 9 icon styles for your button:
Add an optional text label (up to 20 characters) such as "Chat with us", "Get Help", or "Book Now". The label appears next to the icon or in the side tab.
Your settings are automatically reflected in the embed code:
<script src="https://www.harmonix.info/widget.js" data-chatbot-id="YOUR_ID" data-api-key="YOUR_KEY" data-widget-style="side_tab" data-position="right" data-button-icon="sparkles" data-button-text="Chat now"></script>
Copy from the Deploy tab — it always reflects your latest settings.
Track conversations & engagement
The Analytics tab provides insights into:
Manage & convert leads
The Leads section in the dashboard provides a full CRM pipeline to track and manage leads captured by your AI agents.
You can bulk import leads from a CSV file:
Select multiple leads using checkboxes, then use the action bar to:
Automate lead follow-ups
The Automations page lets you create rules that trigger actions when specific events occur.
Automatically send follow-up emails to new leads:
{lead_name}, {company}, {email}, {agent_name}Create agents in minutes
The 5-step guided wizard makes creating a new AI agent easy:
Enter the agent's name, choose a template (Customer Support, Sales, Appointments, or Custom), and select the workspace.
Upload documents or add URLs for the AI to learn from. Supports PDF, Word, and text files.
Define the agent's system prompt, tone, and behavior rules. Set how it should respond and what topics to avoid.
Customize the chat widget's colors, position, and branding to match your website.
Copy the embed code and add it to your website. Your AI agent is now live!
Manage your plan & credits
Go to Settings → Billing to view and manage your subscription.
Need something specific? Use our Custom Plan Builder on the homepage to:
All payments are securely processed through Stripe.
Privacy, terms & more
Review our legal documents to understand how we protect your data and the terms of using Harmonix.
How we collect, use, and protect your data
Terms governing your use of our platform
How we use cookies and similar technologies
Our policies on refunds and cancellations
Guidelines for acceptable platform usage
Common questions about payments & billing
All payments are processed securely by Stripe. For any legal questions, contact legal@harmonix.info.
Our support team is ready to assist you with any questions.
© 2026 Harmonix. All rights reserved.
Powered by VELORIEN GROUP LLC
Version 2.0.0
Social Channel Relay
WhatsApp · Facebook · Instagram · Telegram
Social Media Channels
Connecting Social Media Channels
Your AI agent can automatically respond to messages on WhatsApp, Facebook Messenger, Instagram DMs, and Telegram — converting every conversation into a verified lead.
Channels Overview
Setting Up During Creation (Recommended)
Step 4 of the creation wizard shows all channels. Toggle on any channel and enter credentials. Your agent will be active on those channels immediately after launch.
Setting Up After Creation
How the AI Handles Incoming Messages
When a visitor messages on any connected channel: