Advanced — agents, RAG & fine-tuning

Configure virtual models, build knowledge bases, and fine-tune with LoRA for your use case.

Virtual model & system prompt

1. Create a virtual model with stable slug.
2. Define role, style, and boundaries in prompt.
3. Test with realistic user queries.

Build RAG knowledge base

1. Upload structured docs (FAQs, product docs, SOPs).
2. Check chunking and retrieval quality.
3. Connect KB to your virtual model.

Prepare LoRA fine-tuning

1. Clean training data and remove duplicates.
2. Keep training dialog format consistent.
3. Run first training and review output quality.

Test and deploy

1. Test edge cases and missing-context queries.
2. Compare base vs fine-tuned behavior.
3. Deploy with clear monitoring metrics.

Start path

• Introducing LoRA fine-tuning
• RAG pipeline with your own documents
• Build your first agent with a system prompt