liveBook · Manning

Personalization

As explained in Building Specialized AI Systems

Personalization is adapting a model’s output style, tone, format, or behavior to particular users, brands, or conventions.

Personalization adapts a model’s output style, tone, format, or behavior to maintain a brand voice, follow writing conventions, or reflect user preferences. section 1.5.1

Choose the adaptation strategy by separating knowledge access from behavioral adaptation:

  • Use RAG when external knowledge is required but standard response behavior is acceptable. section 1.5.1
  • Combine RAG with fine-tuning when external knowledge and personalized behavior are both required and sufficient high-quality training data exists. section 1.5.1
  • Use RAG with few-shot learning when personalization is needed but training data is limited. section 1.5.1

How Building Specialized AI Systems covers it

How deep it goes
Defines personalization and compares RAG, fine-tuning, and few-shot alternatives without implementation detail.
Examples
Uses brand voice, writing conventions, and user preferences as personalization targets.
What it assumes
Distinguishes tasks requiring external knowledge from general language, creative writing, and basic reasoning tasks.
What you can do after
Evaluate whether RAG, fine-tuning, or few-shot learning best fits knowledge, personalization, and data constraints.
How it connects
Connects personalization with RAG, fine-tuning, few-shot learning, external knowledge, and training data quality.

Also explained in

Compare how 11 books explain personalization →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Building Specialized AI Systems. The section links point to the parts of the book it draws on.