The Complete Guide To Custom GPTs and Actions

Discover the differences between basic and connected GPTs and how actions connect your custom GPT to external APIs and tools like Kayak and Canva.

Create your first custom GPT using a two-panel interface that defines instructions and actions. Test it in a preview and learn how a GPT is driven by a wrapped prompt.

Explore the GPT builder to create and configure a new custom GPT, name it, set instructions, and craft conversation starters for clear, nontechnical explanations.

Learn to create a custom GPT with the builder, generate short articles, and customize its profile picture by uploading your own image or using DALL-E with specific prompts.

Discover how conversation starters in a custom GPT builder provide quick prompts to start interactions and showcase capabilities, including name, description, and the instruction prompt.

Publish your custom gpt by saving it, choosing access (only me or public), assigning a category, and sharing a published link, with options to unpublish or delete.

Create a plant-identification gpt named green thumb that analyzes user-uploaded photos to identify plants, provide common and scientific names, and deliver a brief, user-friendly explanation.

Learn how to enable DALL-E image generation for a custom GPT and generate logo-ready visuals from prompts, while guiding the model with constraints like no text in logos.

Enable and test web browsing in a custom GPT to retrieve and summarize current stories, as demonstrated with the Happy News Finder that browses Bing for uplifting news.

Enable the code interpreter to let your custom GPT generate downloadable CSV data sets—such as resort attendance, snowfall, and sales—by running Python code, useful for creating realistic teaching datasets.

Explore building a data viz designer: a custom GPT with code interpreter that lets users upload data sets and generate brand-consistent visualizations using a chosen palette and font.

Upload knowledge files, including manuals and PDFs, to a custom GPT to enable a chat bot that answers questions by searching the documents.

Design a custom word cloud generating GPT that accepts text input via upload or typing, uses the code interpreter to create a word cloud graphic, and displays the image.

Learn how GPT actions connect custom GPTs to external APIs and third-party tools, using open API standards and integration methods. See examples like slide makers, stock data, and travel searches.

Learn what an api is and how web apis let a client talk to a server via http or https; explore endpoints, json data, and authentication basics with examples.

Explore the parts of an http request, including the url and http method, and learn how get and post requests differ, plus headers and body and status codes.

Explore http request headers as metadata for api calls, including authentication with api keys and bearer tokens, and learn how accept and content-type headers control response formats.

Understand how post requests send data to an API to create new resources, like to-dos, using headers and body payloads, and how status codes 200, 201, 404 signal results.

Explore the 11 labs text to speech API, authenticate with an API key, and post a request using a voice ID and text to generate audio.

Define actions for a custom GPT using an open API schema. Describe endpoints and paths like /pets with get and post, building human and machine readable API contracts.

Clarify OpenAI and open API terminology, noting that Swagger evolved into the OpenAPI specification and its tools remain branded, while OpenAI's custom GPTs use the open API format.

Develop your first custom GPT action by wiring a simple get request to a cryptocurrency price API, creating an OpenAPI spec, and testing Bitcoin and Ethereum endpoints for live prices.

Explain how API parameters work with a required name in the nationalize API using a Get request, and illustrate writing an open API spec and basic plotting.

Define and document OpenAPI schemas for API response formats, using swagger editor to specify json structures such as count, name, and country probabilities for nationality estimates.

Explore the two data formats used in open api: yaml and json, compare indentation, quotes, braces, and how to convert between them for custom gpt actions.

Configure authentication for actions in a custom gpt, covering api keys, headers, get versus post request flows, and query parameter use, plus proxy api workarounds.

Authenticate with the accuweather api using an api key, locate a city via the locations api autocomplete, then fetch current conditions with the location key for a custom gpt action.

Explore the challenges of using the AccuWeather API inside a custom GPT, including http versus https, API key in the query, and built-in auth limitations, plus a practical workaround.

Learn how built-in auth in custom GPTs handles API keys by using the Postman Echo API to reveal whether keys are sent as headers or query parameters, with practical takeaways.

Proxy API calls through an intermediary API to empower your custom GPT to access APIs while handling API keys. Build a weather API with locations and current weather using NodeJS/express.