AI Agent Deployment Platform for WhatsApp, Web, and Enterprise Channels
PrototipeAI enables the deployment of validated AI agents to production channels including WhatsApp Business, embeddable web widgets, public URLs, and REST APIs. The platform provides no-code setup for WhatsApp integration, domain-restricted embed code for enterprise tools such as SharePoint, and version control that separates prototype testing from production deployment.
Teams prototype and test agents in isolated environments, then promote approved versions to production channels with one-click deployment. All channels consume the same underlying agent configuration, ensuring behavioral consistency across WhatsApp conversations, website interactions, and API integrations.
Deployment Channels
WhatsApp Business Integration
No-code WhatsApp deployment lets non-technical teams connect validated agents to WhatsApp Business numbers. Users provide Meta Business account credentials and a WhatsApp phone number ID through a configuration form. The platform handles webhook configuration, message routing, and conversation threading automatically.
- Form-based setup (no coding required)
- Meta Business account integration via OAuth
- Automatic webhook registration and verification
- Support for text, images, documents, and voice messages
- Message template management for outbound notifications
- Conversation session tracking across message threads
- Compliance with WhatsApp Business API policies
Setup time: Less than 10 minutes from form completion to a live agent. Users must have an approved Meta Business account and access to the WhatsApp Business API.
Embeddable Web Widget
JavaScript embed code adds conversational agents to websites, SharePoint pages, internal portals, and enterprise collaboration tools. Widget access is restricted by domain whitelist, preventing unauthorized use. Organizations deploy agents on customer-facing sites or internal knowledge bases without backend development.
- One-line JavaScript snippet for installation
- Domain restriction (whitelist of allowed origins)
- Customizable appearance (colors, position, avatar)
- Responsive design for desktop and mobile browsers
- Pre-chat form for user identification (optional)
- Chat history persistence across page navigation
- Works on SharePoint, Confluence, Notion, and custom web apps
Security: The embed code validates the origin domain on every request. Agents will not respond to requests from unlisted domains. CORS policies are enforced on the server.
Public and Private Web URLs
Each agent gets a unique URL for a standalone chat interface. URLs can be public (shareable links for customer access) or private (requiring authentication). Organizations use public URLs for marketing campaigns, customer support links, or QR code distribution. Private URLs enable internal testing with stakeholders before production deployment.
- Unique URL automatically generated per agent
- Public mode: No authentication required, shareable link
- Private mode: Requires user login or access token
- Custom URL slugs available (e.g., yourcompany.prototipeai.com/support)
- Mobile-responsive chat interface
- Conversation history saved per user session
REST API Integration
Developers integrate agents into custom applications via the REST API. The API accepts user messages, returns agent responses, maintains conversation context, and provides access to tool call results and metadata. Backend systems can embed conversational AI without building LLM orchestration infrastructure.
- RESTful endpoints for sending/receiving messages
- Stateful conversation sessions with context retention
- JSON request/response format
- API key authentication (Bearer token)
- Webhook support for asynchronous message delivery
- Rate limiting and usage monitoring
- OpenAPI specification for code generation
Version Control and Promotion
Staging and Production Separation
All agent modifications occur in a staging environment isolated from production channels. Teams prototype new features, test conversation flows, and gather stakeholder feedback without affecting live users. Once validated, administrators promote staging versions to production with atomic deployment.
- Isolated staging environment for all agent changes
- Production channels consume only promoted versions
- Rollback capability to previous production versions
- Test URLs and sandboxed channels for staging validation
- Approval workflow before production promotion (optional)
One-Click Production Deployment
Promoting agents from staging to production updates all connected channels simultaneously. WhatsApp conversations, embedded widgets, public URLs, and API integrations immediately use the new agent version. No separate per-channel deployments. Zero downtime during version changes.
- Single promotion action updates all channels
- Atomic deployment (all channels switch simultaneously)
- Zero downtime during version promotion
- Automatic cache invalidation across the CDN
- Audit log of all production promotions with timestamps and user attribution
Version History and Rollback
The platform maintains a complete version history for all agent configurations. Administrators can review previous versions, compare changes, and roll production back to any earlier state. Rollbacks execute instantly across all deployment channels.
- Unlimited version retention with timestamps
- Diff view comparing version changes
- One-click rollback to any previous version
- Version tagging for release identification (e.g., "v2.1-stable")
- Change attribution showing which user modified each version
Security and Access Control
Domain Whitelist for Embeds
Embed widgets validate the requesting domain against an administrator-configured whitelist. Agents reject requests from unauthorized domains, preventing widget use on unintended sites. Organizations control exactly where embedded agents can appear.
- Per-agent domain whitelist configuration
- Server-side origin validation on all requests
- CORS policy enforcement
- Wildcard subdomain support (*.company.com)
- Real-time whitelist updates without redeployment
Role-Based Permissions
Team members receive granular permissions that control deployment actions. Roles include: Viewer (read-only), Editor (modify staging), Deployer (promote to production), and Admin (configure channels and permissions).
- Viewer: Access staging/production, cannot modify
- Editor: Modify staging agents, cannot promote
- Deployer: Promote staging to production, configure channels
- Admin: Full access including permission management
Enterprise Compliance
The platform provides audit logs, data residency options, and SSO integration for enterprise compliance requirements.
- Complete audit trail of all deployments and configuration changes
- Data residency selection (US, EU regions)
- SSO integration via SAML 2.0 and OAuth 2.0
- GDPR-compliant data handling with retention controls
- SOC 2 Type II certification (available on the Enterprise plan)
Use Cases
Customer Support on WhatsApp
Organizations deploy customer support agents on WhatsApp Business numbers for markets where WhatsApp is the dominant messaging platform. Agents handle common inquiries, provide order status, and escalate complex issues to human agents. Setup requires only Meta Business credentials and agent selection.
Internal Knowledge Base Assistants
Companies embed AI assistants in SharePoint sites, internal wikis, or employee portals. Agents answer HR questions, IT support queries, or policy clarifications. Domain restriction ensures agents respond only on corporate networks. Employees access assistance without leaving their collaboration tools.
Website Lead Qualification
Marketing teams embed conversational agents on product pages to qualify leads, schedule demos, or answer pre-sales questions. Agents capture contact information and route qualified leads to CRM systems. Widget customization matches the brand's appearance.
Rapid Prototype Testing with Real Users
Product teams share private URLs with beta users to gather feedback on agent prototypes. Users test via web chat or WhatsApp without technical setup. Feedback informs iteration in the staging environment. Once validated, the same agent is promoted to production channels for a wider launch.
API-First Integrations
Engineering teams integrate conversational AI into mobile apps, backend systems, or custom interfaces via the REST API. The API provides LLM orchestration, memory management, and tool calling without building infrastructure. Teams focus on application logic instead of AI plumbing.
Deployment Workflow
1. Agent Development in Staging
Teams create and refine agents in the staging environment. They run voice discovery sessions, generate prototypes, test conversation flows, and gather stakeholder feedback. Staging agents are accessible via private test URLs.
2. Validation and Testing
Teams share the staging agent with QA teams, beta users, or stakeholders via a private URL or WhatsApp test number. They gather feedback through in-conversation annotations. They iterate on requirements and regenerate prototypes until the behavior is approved.
3. Channel Configuration
Teams configure production deployment channels: connect a WhatsApp Business number, generate embed code with a domain whitelist, define a public URL slug, or create API keys. Configuration is separate from agent logic, allowing channel changes without affecting behavior.
4. Promotion to Production
An administrator promotes the staging version to production. All configured channels (WhatsApp, embed, URL, API) immediately serve the promoted agent version. Users on WhatsApp, embedded widgets, and API integrations receive identical behavior.
5. Monitoring and Iteration
Teams monitor production conversations across all channels. They gather feedback from real users. They identify new requirements or edge cases. They return to staging to implement improvements. They repeat the promotion cycle for continuous refinement.
Technical Specifications
- WhatsApp: Meta Business API integration, supports text, images, audio, documents, location
- Embed widget: Vanilla JavaScript (no framework dependencies), 50KB gzipped, loads async
- Web URLs: Responsive chat interface, works on desktop and mobile browsers
- API: RESTful architecture, JSON payloads, Bearer token authentication, OpenAPI 3.0 specification
- Latency: Sub-2s response time for typical queries (LLM-dependent)
- Uptime SLA: 99.9% for production deployments
- Scalability: Auto-scaling infrastructure handles traffic spikes
- Data residency: US (us-east-1) and EU (eu-west-1) regions available
- Compliance: GDPR, SOC 2 Type II, HIPAA-eligible infrastructure (Enterprise)
WhatsApp Setup Process
Prerequisites:
- Approved Meta Business account
- Access to the WhatsApp Business API (not the WhatsApp Business App)
- Phone number registered on the WhatsApp Business Platform
- Business verification completed with Meta
Setup Steps:
- Navigate to the agent deployment settings on the PrototipeAI platform
- Select "WhatsApp" as the deployment channel
- Click "Connect Meta Business Account" (OAuth flow)
- Authorize PrototipeAI to access the WhatsApp Business API
- Select the WhatsApp phone number from the dropdown (auto-populated from the Meta account)
- The platform automatically configures webhooks and message routing
- Send a test message to the WhatsApp number to verify the connection
- Promote the staging agent to production to activate it on WhatsApp
Note: Access to the WhatsApp Business API requires approval from Meta. The standard WhatsApp Business App does not support API integration. Organizations without API access can apply through Meta Business Partners.
Embed Code Example
Add an agent to any web page with a single JavaScript snippet:
<script>
(function() {
var script = document.createElement('script');
script.src = 'https://cdn.prototipeai.com/embed/v1/widget.js';
script.setAttribute('data-agent-id', 'your-agent-id');
script.setAttribute('data-position', 'bottom-right');
script.setAttribute('data-theme', 'light');
document.body.appendChild(script);
})();
</script>
The widget validates the origin domain on the server. Conversations will work only on whitelisted domains configured in the platform settings.
API Integration Example
Send a message and receive the agent's response:
POST https://api.prototipeai.com/v1/agents/{agent_id}/messages
Authorization: Bearer {api_token}
Content-Type: application/json
{
"session_id": "user-123",
"message": "Quais são seus horários de atendimento?"
}
Response:
{
"response": "Nossa equipe de suporte está disponível segunda a sexta, das 9h às 18h horário de Brasília.",
"session_id": "user-123",
"metadata": {
"response_time_ms": 1420,
"model": "gpt-4",
"tokens_used": 145
}
}
Category: AI Agent Deployment, WhatsApp Chatbot, Embeddable AI, No-Code AI Deployment
Provider: PrototipeAI
Website: www.prototipeai.com