This guide is designed for support teams and content creators to develop clear, structured help desk content that improves the BoldDesk AI Agent’s ability to deliver accurate, relevant responses. Well-crafted content enhances AI 2.0 comprehension, reduces misinterpretation, and elevates the overall customer experience.

This is used to optimize training content, ensuring your AI 2.0 Agent retrieves precise information quickly, providing customers with timely and correct solutions. Better content structure and clarity not only boost AI 2.0 performance but also streamline support operations and increase customer satisfaction.

Help Desk Content Optimization Guide for BoldDesk AI Agent Training

Please follow the guidelines below;

1. Train with High-Quality, Refined Content

Use reliable, well-structured information that reflects real customer needs.

• Use current and accurate KB articles aligned with product features and policies.
• Regularly review and update content after product or policy changes.
• Include FAQs, product documentation, and support workflows.
• Avoid jargon, vague language, and incomplete information.

2. Organize Content by Topic or Department

Structure content to help the AI 2.0 retrieve relevant answers efficiently. Use clear categories and link related articles under each, and group content by functional areas:

• Billing Issues
• Product Setup
• Troubleshooting
• Account Management

3. Use Clear and Hierarchical Headings

Improve AI indexing and user navigation with structured headings.

• Use H1, H2, and H3 to define topic hierarchy.
• Make headings descriptive and keyword-rich.

4. Organize Information with Lists and Tables

Enhance readability and AI parsing with structured formatting.

• Use bullet points for features and numbered lists for steps.
• Present structured data in clean, labelled tables for better AI parsing.

5. Write Short and Focused Paragraphs

Improve clarity and AI 2.0 summarization accuracy.

• Keep sentences short and direct (ideally under 20 words). Avoid complex sentence structures, as they can reduce clarity and lead to misinterpretation by AI.
• Limit each paragraph to a single idea or task. This helps the AI to extract and summarize key information more accurately.

6. Provide Clear, Contextual Links

This helps AI 2.0 to understand topic relationships and improve article relevance.

• Use descriptive anchor text (such as, “Learn how to manage private workspaces”).
• Avoid vague phrases like “click here” or “this article.”

7. Ensure SEO Best Practices

To improve AI and search engine indexing, especially in headings, subheadings, and the opening sentence of each section. Avoid keyword stuffing or repetitive phrasing.

• Use keywords naturally, especially in headings and opening sentences.
• Write a concise meta description with main keywords.
• Add structured metadata (for instance, title tags, descriptions, custom fields) if supported.

Write a clear, concise meta description for each article that includes your main keywords. This helps AI tools understand the topic and improves indexing accuracy.

If supported, add structured metadata (such as title tags, descriptions, and custom fields) to strengthen both search engine and AI recognition.

8. Incorporate Code Snippets Carefully

Help AI 2.0 understand logic and improve developer support content.

• Use clean, properly formatted code with consistent indentation.
• Add meaningful comments to explain logic and usage.
• Use syntax highlighting or language tags (for example, javascript) when your platform supports it.
• Keep code blocks copyable, avoid line numbers or UI styling.

Keep code blocks copyable. Avoid inserting extra formatting (like line numbers or UI styling) that might interfere with reuse.

9. Use Clear and Consistent Terminology

Consistency improves AI understanding and response accuracy.

• Use uniform naming conventions for products and features.
• Stick to one term per concept, unless synonyms are needed.
• Ensure synonyms reflect real user language.

Example:
Instead of “Refer to the image below,” write: “Click the Settings icon in the top-right corner, then select User Management from the dropdown menu.”

10. Describe Visual Content with Textual Instructions

Support AI 2.0 comprehension and accessibility. Please avoid referencing the images directly; instead, write detailed instructions that describe actions, buttons, or sequences. Clearly provide step-by-step textual descriptions of visual action steps depicted in visual media using descriptive language. Use alt text and captions to support both users and AI 2.0 tools.

• Provide step-by-step textual descriptions of visual actions.
• Use alt text and captions to describe images and diagrams.

11. Perform Regular Testing and Updates

Continuously refine your AI 2.0 Agent’s training data to improve performance.

• Use BoldDesk’s testing tools to simulate queries and validate responses.
• Update training content monthly or after major changes.
• Fine-tune based on user feedback and interaction logs.

Rigorously test the AI’s responses and update its training data to minimise errors. Continuously update the AI’s knowledge base with new information and fine-tune models based on user interactions to improve accuracy over time.

12. Mention Ticket and Chat Fields Clearly in AI Knowledge Sources

When creating articles, Q&A entries, or other AI knowledge sources, clearly mention any related ticket fields and chat fields that users may refer to in their questions. This helps AI understand the relationship between field names, field values, and the support scenario being described.

For example, if a workflow, article, or support process depends on a chat field such as Order ID, Customer Type, Issue Category, or Product Name, include the field name and expected values directly in the article content.

Similarly, if a ticket process depends on ticket fields such as Category, Type, Priority, Status, or custom dropdown fields, mention those fields clearly in the relevant article or Q&A entry.

This is especially useful when AI needs to retrieve answers based on field-related values collected during chats or tickets.

Example: If a chat workflow collects the customer’s issue type using a chat field, document it clearly:

• Chat Field: Issue Type
• Possible values: Billing, Technical Support, Account Access, Product Setup
• Usage: AI should use this field to understand the customer’s request category and retrieve the most relevant response from the connected knowledge sources.

If a ticket workflow uses ticket fields for routing or classification, document them clearly:

• Ticket Field: Category
• Possible values: Billing, Feature Request, Bug Report, Login Issue
• Usage: AI should use the field context to understand how the ticket is classified and provide a relevant response.

Question Prompting Best Practices

1. Ask Complete Questions

Do: Frame queries as full questions.
Why: Helps AI understand intent and context.
Example:
“Website hosting pricing plans”
“What are the pricing plans for website hosting services?”

2. Avoid Overly General Queries

Do: Be specific.
Why: Specific questions yield targeted answers.
Example:
“Health insurance”
“What health insurance plans are available for full-time employees?”

3. Use Standard Terminology

Do: Use commonly accepted terms.
Why: Improves compatibility with AI’s training data.
Example:

“Specs of your top-of-the-line tech device”
“What are the specifications of the latest model of the tech device?”

Frequently Asked Questions

1. How often should I update AI 2.0 Agent training content?
   Update AI training content at least monthly and after major product or policy changes.

2. Can I test how AI 2.0 Agent will respond to specific queries?
   Yes. Use BoldDesk testing tools to simulate queries and validate responses.

3. What should I do if AI 2.0 Agent returns incorrect or irrelevant answers?
   Review knowledge sources for gaps, unclear language, or outdated steps. Update the knowledge sources and retrain or reprocess the AI inputs as needed.

4. What formatting improves BoldDesk AI Agent retrieval the most?
   Clear headings (H1/H2/H3), short paragraphs, numbered steps for procedures, and tables for fields and allowed values improve retrieval and parsing.

5. Should KB articles rely on screenshots for instructions?
   No. KB articles should include full textual instructions using UI labels. Screenshots should only support the written steps.

6. Why should KB articles explicitly list chat fields and ticket fields?
   Explicit field names and values help AI 2.0 Agent match user questions to the correct workflow context and retrieve the correct content.

Related Articles

1. BoldDesk AI 2.0 Data Usage, Processing, and Privacy
2. How to Add Knowledge Sources to BoldDesk AI 2.0
3. How to Train AI in BoldDesk Using Knowledge Sources