How the AI agent answers

AnswerRidge's AI agent doesn't make things up and it isn't "trained" on your data in the fine-tuning sense. It uses retrieval-augmented generation (RAG): at answer time it finds the most relevant pieces of your knowledge base and writes a grounded reply from them.

The pipeline

1. Embed the question. The customer's message is converted into a vector.
2. Retrieve. AnswerRidge finds the most semantically similar articles, article chunks and FAQs in your workspace.
3. Compose. The language model writes an answer using only the retrieved content, and cites the source articles.
4. Score. The answer gets a confidence score. Low confidence → escalate instead of guess.

Why this matters

• Always current. Update an article and the AI's answer updates instantly — no retraining.
• Grounded. Answers come from your docs, so they match your policies and voice.
• Auditable. Every answer links back to the source article.

Make it better

The single biggest lever is your knowledge base. More clear, specific articles → better retrieval → better answers. Review knowledge gaps weekly to see exactly what to add.