FeniciaDocs
DocumentaciónAPICambiosSoporte
Buscar⌘K
Contact Center DashboardMessaging InboxAI Assistant

Producto

  • Características
  • Precios
  • Integraciones

Recursos

  • Documentación
  • API
  • Cambios
  • Blog

Empresa

  • Sobre nosotros
  • Contacto
  • Carreras

Legal

  • Privacidad
  • Términos
FeniciaLa plataforma de e-commerce para merchants profesionales

© 2026 Hobbio Inc. Todos los derechos reservados.

AI Assistant

The AI assistant at /contact-center/ai-assistant powers your contact center by generating suggested responses, learning from past conversations, and handling frequent inquiries autonomously.

Response Generation

When an agent is replying in the messaging inbox, the AI assistant analyzes the conversation context and suggests responses:

  • Automatic suggestion: Appears below the reply field with a draft ready to send.
  • Adaptive tone: The AI adjusts tone based on the channel (formal for email, conversational for chat).
  • Order context: If the conversation mentions an order, the AI includes relevant information like status and tracking.

Tip

The agent always has the final say. AI suggestions are editable before sending and are never sent automatically without review.

Training

At /ai-assistant/training you can improve the quality of the assistant's responses:

1

Add conversation examples

Upload model conversations that represent the desired quality of service. The AI learns response patterns, vocabulary, and tone.

2

Define knowledge base

Configure information about your products, return policies, business hours, and frequently asked questions that the AI should know.

3

Set boundaries

Define which types of inquiries the AI can answer autonomously and which should be escalated to a human agent.

Continuous learning

Every time an agent edits an AI suggestion before sending, the system records the correction to improve future responses.

Feedback and Improvement

At /ai-assistant/feedback you will find tools to evaluate and improve the assistant's performance:

MetricDescription
Acceptance ratePercentage of suggestions sent without editing
Edit rateSuggestions modified before sending
Rejection rateSuggestions discarded completely
Post-AI satisfactionCustomer score in AI-assisted conversations

Improvement Cycle

1

Review rejections

Analyze discarded suggestions to identify error patterns or topics where the AI needs more training.

2

Correct examples

Update training examples with the corrections applied by agents.

3

Validate results

Monitor metrics over a week to confirm improvement in the acceptance rate.

Chat Widget

At /contact-center/widget you can configure the chat widget that appears on your store:

  • Appearance: Colors, position, welcome message.
  • Schedules: Define when to show live chat vs. offline form.
  • Autonomous AI: Enable the assistant to respond outside agent hours.
  • Pre-qualification: Configure initial questions to categorize the inquiry before connecting with an agent.

Autonomous mode

When the AI operates without agent supervision (off-hours), make sure escalation boundaries are properly configured to avoid inadequate responses.

Recommended Configuration

For best results from the AI assistant:

  1. Provide at least 50 quality conversation examples.
  2. Update the knowledge base whenever your policies change.
  3. Review feedback metrics weekly.
  4. Enable autonomous mode only after reaching an acceptance rate above 80%.

Tip

Combine the AI assistant with predefined responses to cover both complex inquiries (AI) and standard procedures (templates).

Related: Monitor AI impact on the contact center dashboard.