Master Generative AI: Build AI Assistants with GPTScript

Install and upgrade GPTScript on Mac using brew, uninstall and reinstall to ensure the latest version, and explore models, noting that a key is required to access the different models.

Install GPTScript on Windows using winget in PowerShell, upgrade to version 0.9.4, and restart the terminal to update path. Obtain an OpenAI API key for use in the next lectures.

Learn how to obtain and manage your OpenAI API key for GPTScript, including creating a secret key, enabling two-factor authentication, saving it securely, and setting up billing and models.

Set and verify your open API key on macOS, then run GPT script to list available models, test a hello world script, and explore local tool calls via function calling.

Configure open ai api key in Windows PowerShell, verify it, and run GPTScript examples to explore the default GPT-4 model, script execution, and function callbacks across Windows and macOS.

Explore the structure of a GPT script, with tools as functions using prompts or code, defined by name, description, arguments, and context, including models and max tokens.

Master function calling in GPT Script to connect language models with external tools, enabling AI assistants to call functions like get current weather and to chain or parallelize calls.

Learn to create a GPT script that extracts the top five Ballon d'Or winners, using a dot gpt file, prompts, and caching considerations.

Explore an emerging ui that provides an ide-like debugging and running experience for the top five Ballon d'Or winners script, managed via a local server.

Learn to create a GPTScript that fetches football league standings, extracts team names and points, builds a ranked list, and identifies the most goals and leader's past league wins.

Clarify parameter tool directives and the new type field, showing how args, arg, params, and parameter serve as tool arguments and how type distinguishes tool kinds.

Extend a GPT script to generate a football league HTML page with a styled table featuring sky blue headers, and save data to a local file via an inline tool.

Learn how to create and share context tools in GPT scripts, using static and dynamic football context, to append prompts with up-to-date data and reusable tools.

Define global tools and a global model name in your GPT script to avoid repeating tools and model definitions, then run and write multilingual outputs using the GPT-4 model.

Explore how standard tools and context tools differ in function calling within GPT script workflows. Learn how static and dynamic context augment prompts and update the caller context.

Learn how to manage files with system tools by reading, writing, removing, and finding files via a script that creates and stores a story locally, in a defined order.

Use the system Find tool to traverse folders and locate files by pattern, with Unix-style globbing like star and dot GPT, in current or specified directories.

Install Python on Windows to run Python scripts and apps. Learn how to install it from the Microsoft Store and verify the installation.

Run commands locally with system tool exec, using OS detection for macOS, Windows, or others. A cross-platform script uses callback to run ls -l on macOS/Linux and dir on Windows.

Download content from a URL using HTTP get and HTML to text to feed up-to-date information to your LLM, and explore events extraction, caching, and browser-based browsing limitations.

Learn to use a GitHub-hosted browser tool in GPT Script to crawl websites, extract content, navigate pages, and summarize top links from searches, noting not deterministic results.

Discover how to fetch up-to-date information using a Google-based search tool alongside the LLM, and practice year-by-year Olympic questions to compare web results with model answers.

Explore image generation with the dalle-e model via a gpt script to produce artist-style artworks, saving them as the artist's name and exploring Monet, Cezanne, Picasso, and Kandinsky.

Explore how retrieval augmented generation enhances responses by retrieving external data from a vector database, embedding text into vectors, and augmenting prompts for the large language model.

Manage cached content and tools for GPTScript by locating Mac and Windows cache directories and repos for GitHub tools. Wipe the cache to refresh runtimes and fetch newer versions.

Explore how chats and chat bots work in GPTScript, turning scripts into persistent chat sessions, using reusable context, and managing tools and workspaces for ongoing conversations.

Learn to turn a tool into a pirate chat bot using shot board feature. Set shot to true, instruct the LM to act like a pirate, and finish the chat.

Explore how GPT script chats enable AI assistants that autonomously perceive environments, make decisions, and take actions across domains like healthcare, finance, and customer support.

Build a local cli assistant for github with gpt script, using contexts and tools to execute commands like repo listing, creation, and updates.

Build an API assistant to interact with the DigitalOcean API, manage droplets and databases, authenticate with tokens, and leverage OpenAPI specs and function calling.

Build a local files assistant that searches your file system with plain text queries. Convert queries to operating system commands using GPTScript and a context workspace.

Set up a local Kubernetes cluster with Docker Desktop and use kubectl to check status and resolve TLS certificate issues. Reset, update, and verify cluster readiness and context for development.

Learn to provision a two-node AWS EKS cluster using a generated YAML config, configure the AWS CLI, IAM permissions, and CloudFormation, and manage it with kubectl and Helm.

Test and deploy a Dockerized Python Flask app from AWS ECR by pulling the latest image, running it in daemon mode with port mapping, and validating access inside the container.

Learn how to switch models in GPT script, use OpenAI-compatible APIs or shims for providers like Mistral or Anthropic, and test model compatibility and prompts across providers.

Explore using different GPT models in GPTScript, learn how to specify a model name, view available models, and observe output variability across runs and with caching off.

Explore using an OpenAI-compatible API model from another provider, set an API key, and compare how billing, key setup, and model differences affect responses.