Skip to main content

AI Training (RAG)

The AI Training (RAG) section allows you to give your agent business-specific knowledge by creating a Knowledge Base. This ensures your AI understands your company, products, and customers, not just general information.

What is RAG?​

RAG stands for Retrieval-Augmented Generation an advanced AI framework used by Votel.ai. It works in two stages:

  1. Retrieval: When a user asks a question, the system first searches through your connected data sources (like files and websites) to find the most relevant information.
  2. Generation: The AI then uses that retrieved data to generate a precise, contextual, and human-like response.

This means your agent doesn't rely on assumptions, it answers directly from your verified business knowledge.

Adding Knowledge Sources​

You can train your AI agent using two primary methods:

  1. Upload Files Upload documents such as PDFs, DOCs, or TXT files containing your internal guidelines, product information, or FAQs. Votel will automatically process these files and make their content searchable for your agent.

  2. Add Websites Connect your official website or specific URLs. Votel will crawl the content, scanning text, understanding structure, and indexing relevant data, so your agent can accurately answer customer questions based on your live website.

How Votel Processes Your Files and Websites​

When you upload a file or connect a website, Votel's intelligent processing engine performs a deep scan of the provided content.

Here's what happens behind the scenes:

  • Step 1: Content Extraction: Votel reads through every word and line from your uploaded files or connected web pages.
  • Step 2: Context Understanding: The system identifies key topics, intent, and relationships between pieces of information (e.g., product names, prices, FAQs, policies).
  • Step 3: Logical Memory Creation: Votel then creates a structured, memory-like representation of your content, which allows the agent to recall relevant details during conversations.
  • Step 4: Secure Storage: All processed data is securely stored within Votel's internal memory layer, accessible only when your agent needs to reference it.

So, when a customer asks a question, Votel's AI instantly retrieves the most relevant sentences or facts from this memory, combines them with natural language understanding, and provides an accurate, brand-aligned response, just like a well-trained human representative.

Why It Matters​

With this setup, your agent can:

  • Deliver accurate, data-driven answers to customer queries.
  • Understand your business context instead of relying on general AI knowledge.
  • Maintain consistency across every conversation.
  • Act as a knowledgeable assistant that reflects your brand's real voice and information.

Testing the Knowledge Base​

Once your knowledge sources are added, you can test the agent directly in the platform. Ask questions that relate to your uploaded files or website content to verify how accurately the AI retrieves and uses your data.

💡 Tip: Update your Knowledge Base whenever your business information changes, this ensures your agent always stays current and precise.