liveBook · Manning

Custom GPT

As explained in AI for Everyday IT

A custom GPT is a configurable version of ChatGPT that incorporates user-provided knowledge, instructions, and task-specific behavior.

A custom GPT is a configurable ChatGPT assistant for a specific task. It combines role instructions, uploaded knowledge, conversation starters, and optional capabilities so users do not have to repeat the same context in every conversation. section 2.1.1 section 8.3

Build one around a repeated workplace task:

  • Choose a narrow role, such as a PowerShell assistant, SQL code reviewer, infrastructure documenter, or change-management helper. section 2.1.1
  • Write instructions that define the job, response rules, boundaries, and the context the assistant should use. section 8.3.1
  • Add conversation starters that give users useful ways to begin. section 8.3.1
  • Upload relevant documentation, templates, configuration files, or support procedures as Knowledge. section 2.1.1 section 8.3.1
  • Enable only the capabilities needed, such as web browsing, image generation, or code execution, then save and test the assistant. section 8.3.1

Ticket Analyst

A service desk team can create a custom GPT that analyzes IT service tickets. Its instructions can require categorizing problems and resolutions, examining comments and closure notes, and summarizing trends or anomalies. A user might start with: “Analyze this service ticket for major category, likely cause, resolution, and follow-up action.” section 8.2.3 section 8.3.1

Treat uploaded knowledge as sensitive. Keep assistants private when they contain policies, configurations, feedback, or other proprietary documents, and avoid uploading anything that should not be downloadable by a stranger. Review privacy and model-training settings before using confidential material. section 8.3 section 8.3.1 section 15.1.3

Figure 8.1 Custom GPT interface
figure

Use the interface overview to locate the GPT builder and understand the basic creation experience. section 8.3

How AI for Everyday IT covers it

How deep it goes
Explains role design, instructions, starters, uploaded knowledge, capabilities, saving, and privacy risks without discussing model internals.
Examples
Uses PowerShell, SQL performance tuning, infrastructure documentation, change management, and IT service-ticket analysis as workplace examples.
What it assumes
Assumes access to an OpenAI account, while creating a custom GPT requires a paid Plus account.
What you can do after
Build it from scratch as a focused assistant that analyzes tickets, follows organizational standards, or answers questions from uploaded documents.
How it connects
Connects custom GPTs with ChatGPT, uploaded files, API calls, web browsing, code execution, and service desk operations.

Also explained in

Compare how 2 books explain custom GPT →

Related concepts

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