Initial Setup
This guide walks through the steps to go from a new Agent Red account to handling your first automated customer conversation. Each section describes what happens, what you need to provide, and what configuration options are available.
Agent Red is in early access. If you need onboarding assistance, contact us.
Setup overview
The complete setup path has six stages. Most customers complete the process in under an hour.
1. Account provisioning
When you subscribe to an Agent Red plan (Starter, Professional, or Enterprise), the system provisions a dedicated tenant environment.
What gets created
| Resource | Description |
|---|---|
| Tenant namespace | Isolated environment for your data, agents, and configuration |
| Cosmos DB partition | Your data is stored in a dedicated partition with tenant-level isolation |
| Agent instances | Six AI agents deployed and configured for your account |
| API endpoint | A unique API endpoint for your tenant (e.g., https://agent-red-api-gateway.orangeglacier-f566a4e7.eastus.azurecontainerapps.io/api/v1/tenant-acme) |
| Dashboard access | Web portal for configuration, analytics, and conversation monitoring |
What you need to provide
- Business name — used for tenant identification and branding
- Primary contact email — for account notifications and billing
- Plan selection — Starter ($149/mo), Professional ($399/mo), or Enterprise ($999/mo)
- Billing information — handled through the self-service checkout
First login
Your welcome email contains a Sign in to Dashboard button — but it does not contain your API key. Keys are never sent via email for security. Instead, you sign in using a magic link (passwordless email authentication) and then retrieve your keys from the dashboard.
Step by step:
- Open the welcome email and click Sign in to Dashboard.
- On the login page, click Sign in with magic link.
- Enter the email address used when your account was created.
- Check your inbox for a second email containing a one-time login link (expires in 15 minutes).
- Click the login link — you are now authenticated with an 8-hour session.
- Navigate to Account & billing in the sidebar to view and copy your API key and widget key.
After your first login, you can use either magic link or your API key to sign in. For details on all authentication options, see Securing Agent Red.
Copy your API key to a password manager immediately. The key is displayed on the Account page whenever you are signed in, but storing it in a password manager avoids repeating the magic link flow every time.
2. API key configuration
Each tenant receives API keys for authenticating requests to the Agent Red API. Your welcome email includes a link to the admin dashboard where you can view and manage your keys.
API key security
- API keys are stored in Azure Key Vault with Managed Identity access
- Each key is scoped to a single tenant — it cannot access other tenants' data
- Keys can be rotated from the dashboard without downtime
- All API calls require the key in the
X-API-Keyheader (for tenant/user keys) orX-Widget-Keyheader (for widget keys)
Authentication format
Agent Red uses two authentication headers depending on the key type:
Tenant or user API key (prefix ar_user_):
curl -X POST https://agent-red-api-gateway.orangeglacier-f566a4e7.eastus.azurecontainerapps.io/api/chat/conversations \
-H "X-API-Key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"message": "Where is my order #12345?"}'
Widget key (prefix pk_live_) — used by the embedded chat widget:
curl -X POST https://agent-red-api-gateway.orangeglacier-f566a4e7.eastus.azurecontainerapps.io/api/chat/conversations \
-H "X-Widget-Key: YOUR_WIDGET_KEY" \
-H "Content-Type: application/json" \
-d '{"message": "Where is my order #12345?"}'
Widget keys are restricted to chat and configuration endpoints only (/api/chat/, /ws/chat/, /api/config).
3. Connect data sources
Agent Red needs access to your product and customer data to provide accurate, context-aware responses. The primary integration is Shopify.
Shopify integration (included in all plans)
The Shopify integration syncs your product catalog, order data, and customer profiles into the knowledge base. Agent Red uses OAuth for secure, permissioned access.
Required Shopify permissions:
| Scope | Purpose |
|---|---|
read_products | Access product names, descriptions, prices, and availability |
read_orders | Look up order status, tracking numbers, and delivery dates |
read_customers | Identify returning customers and access order history |
read_inventory | Check stock levels for availability questions |
Sync behavior:
- Initial sync — full catalog import when you connect your store (typically 5–15 minutes)
- Ongoing sync — webhook-driven updates whenever products, orders, or inventory change
- Sync frequency — real-time for order updates; every 15 minutes for catalog changes
4. Knowledge base setup
The knowledge base is what makes Agent Red's responses accurate and specific to your business. It combines three data sources:
Product catalog (automatic)
The product catalog syncs from Shopify automatically. No manual configuration is required. The knowledge retrieval agent uses this data to answer product questions, check availability, and provide pricing information.
FAQ entries (manual)
FAQ entries are question-and-answer pairs that you create in the dashboard. They cover common questions that are specific to your business and not covered by product data alone.
Examples of effective FAQ entries:
- "What is your return policy?" → Your specific return policy text
- "Do you offer gift wrapping?" → Your gift wrapping availability and pricing
- "How long does shipping take?" → Your shipping timeframes by region
Best practices:
- Write answers in the tone you want the AI to use with customers
- Cover the 20–30 most common questions first (analytics data helps identify these after launch)
- Update entries when policies change — the knowledge base re-indexes in minutes
Policy documents (manual)
Policy documents are longer-form content (return policies, warranty terms, shipping rules) that the AI references when answering policy-related questions. Upload these as text in the dashboard. The system chunks and indexes them automatically.
5. AI Configuration wizard
When you first log in to the admin console and have not yet activated a configuration, Agent Red presents a three-step setup wizard. The wizard guides you through the essential configuration in order.
Step 1 — Connect your store
Link your Shopify storefront so Agent Red can automatically import product catalog, policies, and page content.
- Click Connect Shopify in the wizard.
- Authorize the required permissions (read_products, read_orders, read_customers, read_inventory).
- Wait for the initial sync to complete (typically 5–15 minutes depending on catalog size).
- The wizard shows a summary of imported content — product count, policy pages found, and custom pages.
If you are using Agent Red without Shopify (standalone mode), click Skip to proceed to Step 2 and add knowledge base content manually.
Step 2 — Review your knowledge base
Browse the articles that were imported from your storefront and optionally apply an industry template.
- Review imported articles — they appear as drafts. Edit titles and content for clarity.
- Optionally click Apply template to add industry-specific FAQ articles (11 categories available — Electronics, Apparel, Beauty, Food, Health, Home, Jewelry, Pet, Sports, Toys, General).
- Publish the articles you want the AI to use. Only published articles are searchable.
- Add any additional FAQ entries or policy documents that were not covered by the import.
Spend 10–15 minutes reviewing imported articles. Auto-imported content may need editing for tone, accuracy, or completeness. This step has the single biggest impact on response quality.
Step 3 — Configure your agent
Review AI-generated configuration suggestions and customize your agent's personality and behavior.
- Brand name — pre-filled from your store name. Edit if needed.
- Brand voice — AI analyzes your content and suggests a personality description. Accept or write your own.
- Formality — choose Casual, Professional, or Formal.
- Response length — choose Concise, Moderate, or Detailed.
- Policies — review the return window (days), refund policy, and shipping policy. AI suggestions appear as badges — click to preview and apply.
- Escalation threshold — a slider from 0.0 (aggressive AI) to 1.0 (conservative). Start with your tier's default.
- Custom instructions — optional free-form instructions for the AI (e.g., "Always mention our free shipping over $50").
Configuration inputs with AI suggestions show a colored badge. Click the badge to preview the suggestion, then click Apply to accept it.
Activate
After completing all three steps, click Activate to publish your configuration and go live. The chat widget becomes active on your storefront.
The onboarding wizard only appears before your first activation. After that, you manage all settings through the standard admin pages (Configuration, Knowledge Base, Widget, etc.).
6. Go live
Before enabling Agent Red for live customer traffic, review these configuration options.
Pre-launch checklist
- Knowledge base populated — product catalog synced, FAQ entries created, policy documents uploaded
- Escalation rules configured — define which situations should route to a human agent (e.g., refund requests over a dollar threshold, complaints, VIP customers)
- Brand voice configured — set the tone, greeting style, and sign-off that match your brand
- Test conversations completed — run 10–20 test conversations covering common intents to verify response quality
- Monitoring configured — set up alert thresholds for escalation rate, response latency, and error rate
Deployment options
| Option | Description | Recommended for |
|---|---|---|
| Shadow mode | Agent Red processes messages but does not deliver responses. Human agents see AI-suggested replies for review. | First 1–2 weeks. Build confidence before full automation. |
| Hybrid mode | Agent Red handles low-risk intents (product questions, order status) automatically. Complex intents route to humans. | Ongoing. Most customers use this mode. |
| Full automation | Agent Red handles all intents, with escalation as the safety net. | High-volume stores with well-tuned knowledge bases. |
Monitoring after launch
The Agent Red dashboard provides real-time visibility into:
- Conversation volume — messages handled per hour/day/week
- Automation rate — percentage of conversations resolved without human intervention
- Escalation rate — percentage routed to human agents (target: under 15%)
- Response quality scores — Critic validation pass rate
- Customer satisfaction — inferred from conversation outcomes and explicit feedback
Next steps
- Shopify Integration — Detailed Shopify connection guide.
- How It Works — Deep dive into the agent pipeline and communication protocols.
© 2026 Remaker Digital, a DBA of VanDusen & Palmeter, LLC. All rights reserved.
Was this page helpful?