AI tools decision matrix

Adobe Firefly is an image-based tool that enables users to generate images and text effects simply by typing keywords or descriptions. It is trained on stock images, openly licensed, and public domain content, making it ideal for creative tasks. Users can create graphics, designs, and text effects directly through Adobe Creative Cloud applications, without needing advanced design skills.

Amazon Q allows users to create their own AI-driven chat applications.  Amazon Q is a no-code solution, enabling users to leverage AI capabilities without extensive coding knowledge. This tool is designed for users seeking to integrate AI chat functionalities into their services or applications, providing them with the ability to customize interactions based on specific needs or tasks and integrate with a library of enterprise and productivity applications.

Apigee OpenAI APIs provide developers with access to OpenAI GPT large language models via API, facilitating the creation of code-based applications and offering HUIT consolidated billing.  With support for data classified at Level 3 and below, it caters to developers seeking to integrate advanced AI functionalities into their projects.

AWS Bedrock provides developer access to a variety of third-party large language models via API. It enables developers to integrate advanced AI capabilities into their applications, and is ideal for creating applications that require sophisticated natural language understanding or generation, or image processing.

AWS SageMaker is a managed service that offers developers and data scientists the ability to build, train, and deploy machine learning (ML) models quickly. It provides easy access to ML model training and deployment, supporting a wide range of machine learning frameworks.  SageMaker provides tools for each step of model development, from preparation and analysis of data to model building and tuning. It’s ideal for those looking to leverage ML capabilities without the need for extensive infrastructure management.

Azure AI Studio enables developers to create generative AI applications using the Azure suite of tools and models. You can explore and interact with large language models, build and test prompt applications with prompt flow, fine-tune models for your specific needs, and deploy and monitor AI applications.

Azure OpenAI APIs provides secure access to various OpenAI models for developer integration into AI-related projects.  Access requires the creation of an Azure account and offers benefits such as negotiated rates and simplified billing through Harvard’s enterprise agreement with Azure.

ChatGPT Enterprise GPT is a versatile custom AI chatbot capable of generating text, images, and executable code based on user inputs and a knowledge store of uploaded content. It offers a highly customizable experience, allowing users to tailor responses for various use cases including administrative tasks, creative content creation, and analytical work. Designed to handle data classified at Level 3 and below, it is  suitable for end-users seeking a no-code solution for a wide range of projects.

OpenAI Copilot for Microsoft 365 is an AI add-on designed to enhance productivity within Microsoft’s suite of applications. It assists users by generating administrative, creative, and analytical content, directly within the Microsoft 365 ecosystem. Users can expect to produce a wide range of outputs, from email drafts and document content to data analyses and creative writing.  Copilot for M365 adapts to the context provided by the user’s input in their content, providing a seamless integration into daily workflows.

Copilot Studio is a no-code platform for creating custom AI chatbots.  Users can leverage Copilot Studio to develop tailored chatbot interactions for administrative, creative, or other specific tasks, using OneDrive, sharepoint, or web-based content.

Gemini for Google Workspace, formerly Duet AI, is an AI add-on for Google Apps designed to assist users in creating administrative, creative, and analytical content within Google’s productivity applications. It facilitates efficient workflows and enables end-users to produce content, analyze data, and generate creative solutions directly within Google’s ecosystem.

GitHub Copilot is an AI coding assistant that writes code from natural language or examples. It creates code from natural language, suggests code completions, integrates with various integrated developmnent environments (IDEs) and natively supports GitHub.

Google Vertex AI offers developer access to Google’s advanced AI models, enabling the creation and deployment of machine learning projects. It’s designed for developers looking to integrate Google’s AI capabilities into  applications, facilitating tasks such as natural language processing, computer vision, and predictive analytics. Suitable for code-level tasks, Vertex AI supports a range of AI and machine learning needs, from model training to deployment.

The Harvard AI Sandbox offers a secure environment for experimenting with Generative AI, mitigating security and privacy risks. Users can interact with multiple Large Language Models (LLMs) to produce text-based outputs suitable for administrative, creative, or analytical tasks. It supports data up to Level 3 and provides limited file upload functionality.  The Harvard AI Sandbox is available at no-cost to faculty, staff, and students enrolled in a qualified course.

Microsoft Copilot, formerly known as Bing Chat, is a versatile AI chatbot that enables users to generate both text-based content and images. It’s designed for a range of tasks, including administrative and creative applications. Accessible through Harvard M365 accounts, Microsoft Copilot supports Level 3 data, is suitable for drafting documents and creating images, all without any associated costs and fully supported by HUIT.  Microsoft Copilot is available at no-cost to faculty, staff, and students.

OpenAI APIs offer developer access to advanced GPT large language models via API, enabling the creation of a wide range of applications from simple chatbots to complex analytical tools. OpenAI APIs support data classified up to Level 1 (or Level 3 under Harvard’s enterprise agreement with OpenAI) and allow developers to incorporate cutting-edge AI capabilities into their projects.

OpenAI ChatGPT Enterprise is an advanced version of the ChatGPT tool, designed for producing a wide array of outputs including text, images, and executable code from inputs that include text, files, and images.  ChatGPT Enterprise provides for use of Level 3 data. ChatGPT Enterprise facilitates a broad spectrum of tasks for end-users, ranging from administrative and creative to analytical, supported through both text inputs and file uploads. 

Access to open source AI tools and LLMs on dedicated GPU hardware.