The answer
No. An existing assistant model may already perform the task well with clear instructions and relevant context. Fine-tuning should address a demonstrated behavior gap.
What are the main stages involved in building an AI assistant?
Is supervised fine-tuning always required?
More FAQs in What are the main stages involved in building an AI assistant?
No. Most application teams can start with an existing model and focus on prompts, retrieval, tools, and evaluation. Pre-training is a separate, resource-intensive undertaking.
Evaluation should begin before customization and continue through development and deployment. Early baselines help determine whether later training or application changes actually improve behavior.
Still have questions?
Browse all our FAQs or reach out to our support team
Want to upskill yourself?
Your learning library
Small steps. Big possibilities.
Find your next course. Build something great.