Custom data

As explained in The Art of AI Product Development

Data tailored to a particular application, user population, or domain for adapting or evaluating a model.

Custom data gives a language-model application the domain knowledge, user context, and current information that generic outputs often miss. It may appear as few-shot examples, a retrieval database, or fine-tuning data, and it must remain representative and up to date. section 5.4.5 section 7.0

A marketing assistant with proprietary knowledge

A B2B SaaS marketing team finds that generated content lacks specificity because relevant insights are scattered across Confluence, Salesforce, Google Drive, and other internal sources. Retrieval-augmented generation can retrieve that information at response time, avoiding constant retraining while keeping content aligned with the company’s audience and expertise. section 7.1.2

Treat custom data as an operating practice rather than a one-time upload:

• Review model failures and add correcting examples to the dataset. section 5.4.5
• Monitor whether data still represents changing trends, users, and expectations. section 5.4.5
• Automate data monitoring, error tracking, evaluation, and updates where practical. section 5.4.5

Figure 7.1 From semantic search to a full RAG system

Use the progression from semantic search to a full RAG system to connect custom knowledge with generated responses. section 7.0

How The Art of AI Product Development covers it

How deep it goes

Covers representative-data refinement, corrective examples, RAG, and fine-tuning, but stops short of implementation detail.

Examples

Uses internal wikis, knowledge bases, Confluence, Salesforce, Google Drive, few-shot examples, and failure corrections.

What it assumes

Assumes a language-model application and teaches why generic, outdated outputs require domain-specific, current information.

What you can do after

Apply it through RAG, few-shot examples, fine-tuning, and iterative error-driven dataset updates.

How it connects

Connects to retrieval-augmented generation, semantic search, fine-tuning, prompting, DataOps, and MLOps.

liveBookmanning.com

Read The Art of AI Product Development on liveBook Get the book at manning.com

Related concepts

• retrieval-augmented generationused by · Retrieval-augmented generation is a technique that retrieves relevant information from external sources and supplies it as context to a generative model when producing an output.
• language modelconfigures · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
• sensitive informationuses · Sensitive information is information requiring protection because its unauthorized access, disclosure, alteration, or misuse could cause harm or violate obligations.
• datakind of · Data is represented information, observations, measurements, or other values that can be stored, processed, transmitted, analyzed, or interpreted.

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