Training Your AI with Common Questions

Build your AI's knowledge base with common questions and answers specific to your business. A well-trained FAQ knowledge base means your AI handles 80% of caller questions without needing to transfer, saving time and impressing callers.

🎯 What You Will Learn

Structuring your knowledge base with categories and hierarchies
Writing effective Q&A pairs that handle question variations naturally
Using call logs to identify gaps and continuously improve responses
Testing and validating AI responses before deploying changes

⚠️ Prerequisites

A list of your top 20 most frequently asked questions
Accurate answers for each question (verify with your team)
Access to recent call transcripts to identify additional common questions
FORAI account with basic setup complete

🚀 Step-by-Step Guide

Access the Knowledge Base editor

Navigate to Settings > Knowledge Base. The editor shows your current Q&A entries organized by category. If this is your first time, you'll see a mostly empty workspace with a 'Quick Start' wizard that walks you through adding your first entries.

Settings - Knowledge Base Editor

Create categories for your FAQ topics

Organize your knowledge base into logical categories: Pricing & Billing, Services, Hours & Location, Policies, Technical Support, etc. Categories help the AI match questions faster and let you manage large knowledge bases efficiently.

Write your first Q&A pairs

For each entry, write the Question (how a caller might phrase it) and the Answer (what the AI should say). Write the answer in the same conversational tone your AI uses. Keep answers under 3 sentences for phone delivery -- callers can't scroll back.

Add question variations

For each Q&A, add 3-5 alternative phrasings. For 'What are your hours?', also add 'When do you open?', 'Are you open on weekends?', 'What time do you close?'. More variations = better matching accuracy.

Q&A Entry with Variations

Include contextual follow-up information

For complex topics, add follow-up context. If someone asks about pricing, the AI can answer and then offer: 'Would you like me to text you our full pricing guide?' This creates a natural conversation flow rather than dead-end answers.

Test AI responses in the simulator

Use the 'Test Knowledge Base' tool to ask questions as a caller would. Vary your phrasing to test matching accuracy. The simulator shows which Q&A entry matched and the confidence score. Entries scoring below 80% need more variations.

Knowledge Base Test Simulator

Review call logs for gaps

Go to Analytics > Call Transcripts and filter for calls where the AI said 'I don't have that information' or transferred to a human. These are your knowledge gaps. Add new Q&A entries for each recurring gap.

Schedule regular knowledge base updates

Set a calendar reminder to review and update your knowledge base monthly. As your business changes (new services, price changes, policy updates), the knowledge base must stay current. Stale information is worse than no information.

💡 Pro Tips

Start with your top 20 questions -- this handles roughly 80% of caller inquiries. You can always expand later. Quality over quantity.

Write answers using everyday language, not industry jargon. If a caller asks 'How much does it cost?', don't respond with 'Our fee schedule is commensurate with the standard of care.' Say 'A first visit is typically $150.'

Use the 'Suggested Questions' feature in Analytics, which uses AI to identify questions callers frequently ask that aren't in your knowledge base yet.

Review the 'Low Confidence Matches' report weekly. These are calls where the AI matched a Q&A entry but wasn't very confident. They often reveal phrasings you haven't accounted for.

⚠️ Important Warning

Always test your configuration with the built-in simulator before deploying to live callers. Changes to voice settings, routing rules, and workflows take effect within 60 seconds. Monitor the first 5-10 live calls after any change and revert immediately if you notice issues.

⚙️ Troubleshooting