Microsoft and Khan Academy partnered to provide free access to Khanmigo for Teachers in English in more than 60 countries/regions. Khanmigo for Teachers is an AI-powered teaching assistant powered by Azure AI. It streamlines class preparation tasks and gives educators more time and energy to spend with learners.
Khanmigo offers:
Standards-aligned lesson planning tied to Khan Academy’s world-class content library
An on-demand summary of recent learner work so educators can quickly assess progress and identify areas where learners need more support
Khanmigo-crafted learning objectives, rubrics, and exit tickets
As a planning assistant and instructional collaborator, Khanmigo uses Khan Academy content to simplify AI for educators. No prompting is required. Khanmigo helps:
Create engaging lesson hooks
Provide insights on learner performance
Recommend assignments
Offer support for refreshing knowledge on a topic
The Khanmigo tools page includes over 25 educator-focused Khanmigo activities in five categories:
Plan
Create
Differentiate
Support
Learn
Within the Plan category, educators can:
Review learners’ performance and trends with Khan academy lessons
Craft engaging prompts to stimulate meaningful class discussions
Create quick end-of-lesson assessments to check learner understanding
Develop clear, measurable learning objectives to guide instruction
Plan compelling lesson starters to engage learners
Create structured, detailed lesson plans tailored to curriculum and learners’ needs
Receive recommendations on what learners need to work on next
Within the Create category, educators can:
Create questions on a variety of topics to export to Blooket
Produce engaging newsletters to keep guardians informed about class activities
Generate concise, easy-to-follow instructions for assignments and activities
Transform memorable class moments into a creative poetic recap
Create informational text for a variety of topics
Create personalized letters of recommendation
Create multiple-choice quizzes on a variety of topics
Minecraft Education’s AI Foundations offers a set of accessible, engaging materials for building AI literacy with Minecraft. The program is designed to empower learners, educators, and families with a fundamental understanding of how AI works and how to use AI tools responsibly. A series of animated videos teaches learners the basics of AI. Then, learners play through real-world scenarios in immersive Minecraft worlds.
Minecraft Education’s AI literacy worlds help learners:
Discover how AI can help us solve real-world problems
Demystify the risks and opportunities of AI
Build confidence through game-based learning
The curriculum includes the AI Adventurers video series. With three short videos, learners journey alongside two curious companions and explore the world of AI. The series is produced in partnership with Microsoft’s Democracy Forward initiative and includes:
Class-ready teaching materials
A guide for guardians
Learners can then explore the principles of responsible AI and discover ways AI can help solve real-world problems through a set of Minecraft coding lessons.
Fantastic Fairgrounds: Build AI literacy and learn about history, ethics, and careers in AI.
AI for Earth: Learn the principles of AI by exploring the use of AI for preserving wildlife and ecosystems, helping people in remote areas, and researching climate change.
Hour of Code: Generation AI: Venture through time to create helpful AI-powered inventions, using problem solving, creativity, and computational thinking along with the principles of responsible AI.
Hour of Code: AI for Good: Learn coding basics and explore a real-world example of artificial intelligence by collecting data about forest fires.
Coding with Minecraft: AI: Explore the connection between AI and computer science.
The Investigations: Improve information literacy skills by evaluating biases, decoding clues, and uncovering the truth.
Educators can also use the Minecraft Prompt Lab to learn how to use Copilot Chat with sample prompts. With Minecraft Prompt Lab for Educators, educators:
Explore the basics of writing a good educational prompt
Learn how to use Copilot Chat to create engaging game-based STEM lesson plans for Minecraft Education
Learn how to use Copilot Chat to create assessment materials personalized to learners’ needs, curriculum standards, and educational goals
With Minecraft Education’s AI literacy solutions, families can explore AI together with discussion guides and fun gameplay, and educators can use the AI Foundations program to:
Microsoft Teams for Education includes free AI-powered features that save educators valuable time.
AI-powered features in Teams
In class teams, educators can use AI to draft:
Rubrics
Assignment instructions
Learning objectives
Classwork modules
AI tools in Teams Assignments
When educators create an assignment in Teams, AI can offer suggestions and tips on:
Assignment instructions
Learning objectives
Rubrics
And more
AI offers multiple suggestions on how to enrich the content. After the content is created, educators can make edits or generate different options.
Generate Classwork modules with AI
Educators can also use the Classwork AI module generation feature to save time and find inspiration for designing their Classwork modules. Educators can use AI to generate complete modules and descriptions by providing some basic information about their class. It can help educators create course outlines from scratch or enhance existing courses with more modules. Educators can also modify the description length and style to adjust and regenerate modules.
Learn more about using Microsoft Teams for Education by completing the Master Microsoft Teams for any learning environment learning path in the Microsoft Learn Educator Center.
Learning Accelerators is a category of AI-powered learning tools included in Microsoft 365 for Education. They’re accessible in Teams assignments and the Assignments LTI (Learning Tools Interoperability). Learning Accelerator tools are designed to help educators streamline the creation, review, and analysis of practice assignments for valuable skills. They also provide learners with real-time coaching along the way to help them catch up, keep up, and get ahead. In short, Learning Accelerators help educators unlock the full potential of every learner. These tools support foundational and future-ready skills. They help educators give individual learners more opportunities to learn, practice, and receive targeted coaching instantly in an inclusive environment.
Learning Accelerators cover two core areas:
Foundational skills, including reading, numeracy, and emotional wellbeing
Future-ready skills, including digital literacy (web search) and presentation (public speaking)
With these powerful tools in Microsoft 365 for Education, learners can start practicing right away and educators can start gathering data to help learners build skills.
Reading Progress
Reading Progress tracks learners’ reading skills, gives educators actionable insights quickly, and focuses learners on specific areas for improvement. By streamlining the reading assignment creation, review, and analysis process, educators can spend more of their time on active instruction.
Reading Progress is available within Microsoft Teams for Education. To create Reading Progress assignments, educators can:
Upload their own reading passages
Select a reading passage provided by ReadWorks
Create AI-generated reading passages.
Educators can also use AI-generated comprehension questions for reading passages they assign. Learn more about Reading Progress by completing the Support reading fluency practice with Reading Progress module in the Microsoft Learn Educator Center.
Reading Coach
Reading Coach provides personalized, engaging, consistent, and independent reading fluency practice. It uses artificial intelligence and built-in fluency detection to personalize reading content, using words learners struggle with.
Reading Coach features:
Safe, one-of-a-kind AI-generated stories
Access to the fully accessible Immersive Reader
Targeted practice on challenging words
Rewards that keep learners inspired and motivated
Reading Coach develops a love for reading by giving learners the option to create their own reading materials. It offers three reading modes:
Create a story using AI
Read a passage from the library
Add your own passage
Create a story using AI helps the reader own their reading by giving them the choice of a character, setting, and reading level to create a unique story with AI. Learners can create stories with characters and settings of their choosing from a curated collection of appropriate and engaging options. As learners read a story aloud, speech-to-text artificial intelligence:
Analyzes their reading fluency
Detects words they found challenging
Records the reader’s:
Accuracy
Words per minute
Time spent reading
Reading Coach provides targeted word practice after each session with tools that help learners gain confidence by working through challenging words.
Reading Coach is available for free as a Windows app and a web app to use in class or at home with a Microsoft account. Access it online at https://coach.microsoft.com. Learn more about Reading Coach by completing the Build reading fluency with Reading Coach module in the Microsoft Learn Educator Center.
Search Coach and Search Progress
Search Coach and Search Progress are free tools built into Teams for Education that empower learners to think critically and search online with confidence, which is more important than ever in the age of AI.
Search Coach provides real-time coaching to learners on how to form effective queries and identify and review reliable resources in a secure, ad-free online search environment.
Search Progress in Teams Assignments helps to scaffold the research process, with reflection opportunities that prompt learners to “show their work” as they collect sources. It’s designed to make it easier for educators to build information literacy into any research assignment, no matter the subject.
While Search Coach is ideal for real-time lessons in class and ad-hoc, learner-driven searching, Search Progress adds more structure by integrating it right into Teams assignments. With Search Progress, educators have a detailed view of learners’ research processes and critical thinking over any research project.
Learn more about information literacy fundamentals and using Search Coach and Search Progress by:
Speaker Progress helps learners develop confidence in their presentation skills. It also helps reduce anxiety by providing AI-powered real-time coaching and feedback on public speaking skills like pace, pitch, filler words, and more.
Speaker Progress can save educators time and create more opportunities for independent practice for in-class presentations. Speaker Progress is available in Microsoft Teams for Education. Within Teams Assignments, educators can create new Speaker Progress activities. Learners then practice presenting and receive real-time feedback while they present. They also receive a rehearsal report at the end of each attempt with their top strengths and top opportunities for improvement. Educators can then review learners’ speeches, review learners’ rehearsal reports, and then track progress on key presentation skills over time.
Speaker Coach is a free tool built into Microsoft PowerPoint and Teams meetings. In PowerPoint, learners can select Rehearse with Coach to have a private rehearsal session for their presentation at any time. In Teams meetings, learners can use Speaker Coach to develop their interaction skills.
Microsoft 365 Copilot Chat is an AI tool that can create content like text, images, poems, and code. Copilot Chat can be a versatile AI assistant for education. Copilot Chat offers AI-powered web chat using the latest AI models, enterprise data protection, and citations for the sources it uses. It generates custom answers using current web data, and educators can ask follow-up questions to get the answers they need. Plus, responses come with sources, so educators and learners know exactly where the information came from and can review the citations for generated responses.
Copilot Chat builds on the existing Bing search experience to provide a new way to find information, generate content, and accelerate tasks. Instead of receiving a list of relevant sources as with traditional Bing searches, Copilot consolidates sources from across the web to give a single, summarized answer.
The Copilot Chat experience is a culmination of four technical breakthroughs:
Copilot Chat runs on new, next-generation large language models (LLMs), like GPT-4o, which take key learnings and advancements from applications like ChatGPT.
Microsoft’s Prometheus model is a proprietary way of working with OpenAI that uses the power of new LLMs to produce more relevant, timely, and targeted results with improved safety.
Copilot Chat uses this AI model in its core search ranking engine to ensure that basic search queries are more accurate, relevant, and current.
Microsoft developed a new way for users to interact with search, browser, and chat by pulling them into a unified experience to unlock a new way to interact with the web.
For educators and learners who are signed in with their school account, enterprise data protection offers an ad-free interface tailored for educational environments. With enterprise data protection, the chat data educators and learners enter in Copilot Chat:
Is protected by encryption, logical isolation, and compliance with privacy laws like GDPR
Is safeguarded from external access
Isn’t used to train the underlying AI models
Additionally, enterprise data protection in Copilot Chat:
Safeguards against AI-focused risks like harmful content and prompt injections
Applies a school’s existing access controls and policies to Copilot Chat to ensure prompts and responses are logged, retained, and available for audit, eDiscovery, and advanced Microsoft Purview capabilities
Note
Educators and learners can confirm they’re signed in with their school account when the shield icon appears in the upper-right corner of the Microsoft 365 Copilot Chat page.
To get the best results, educators must learn how to speak the language of AI with successful prompting. Crafting effective prompts to generate responses isn’t an exact science; iteration is key to guiding the model to the results you want. Microsoft recommends including the following parameters to craft an effective prompt:
Goal: Define the end goal or action that you want Copilot Chat to provide
Context: Explain why you need this information or how you’ll use it
Expectations: Share the format or target audience you want the response tailored to
Source: Identify known information or data sources Copilot Chat should use
It’s important to remember that even with the best prompt, AI can make mistakes. With every result, educators should:
Use their expertise and judgment
Double-check facts and linked sources
Report any inappropriate and inaccurate content
Microsoft 365 Copilot Chat is available to education administration, staff, faculty, and higher education learners aged 18 and older at no additional cost. It’s a valuable tool for everyone that offers new opportunities for content creation, personalization, efficiency, and innovation.
Copilot Chat can help school administrators:
Save time by expediting routine tasks like developing transition materials and planning field trips and assemblies, freeing up time for strategic planning
Enhance communication by generating drafts and personalized communications that ensure consistency and clarity
Organize class materials and resources by generating procedures and protocols for checking out and returning technology and other shared resources
Develop professional development resources for educators based on their individual needs
Generate images to enhance presentations
Copilot Chat can help educators:
Create personalized lesson plans and materials tailored to individual learner needs
Focus their time on teaching by providing detailed personalized feedback for learners
Continue their professional growth by finding professional development resources
Enhance engagement by designing more interactive lessons
Generate images to enhance learning by adding images that illustrate abstract ideas, historical events, scientific processes, and more
Copilot Chat can help learners:
Develop future-ready skills by gaining familiarity with advanced technologies
Receive support and guidance with study tips and help studying for assessments by generating study guides and practice quizzes
Enjoy an enhanced learning experience with personalized feedback, accessible learning materials, and the ability to find supplemental resources to support understanding
Spark creativity and find inspiration for creative works or images
Enhance presentations and other works by generating images
According to a Pew Research survey published in April 2024, 84% of educators “say there’s not enough time during their regular work hours to do tasks like grading, planning, paperwork, and answering work emails.”1 81% of those educators reasoned that they simply have too much work to complete it during their regular work hours.
At the same time, developments in artificial intelligence (AI) offer opportunities to reshape the way schools approach creation, problem solving, learning, and communication. Critical thinking and metacognitive skills are more important than ever as AI advances.
With the ever-growing list of demands and improvements in AI, educators and administrators should explore how AI can help them save time and focus more on teaching and addressing learners’ needs. Microsoft offers a variety of free AI tools that can:
Support personalized learning
Provide valuable insights and detailed feedback
Help with repetitive tasks
These free AI tools help educators and administrators bring new opportunities to schools. Educators and administrators can use Microsoft’s free AI tools to:
Work efficiently and creatively with Microsoft 365 Copilot Chat
Improve coding skills with GitHub Copilot
Unlock learners’ full potential with Microsoft Learning Accelerators
Save time with Microsoft Teams for Education
Streamline day-to-day prep and meet the needs of learners with Khanmigo for Teachers
Build AI literacy and digital citizenship with Minecraft Education’s AI Foundations
When educators integrate these AI tools, they can create engaging lesson plans, provide detailed feedback, and develop future-ready skills in their learners. Educators can use these free AI tools to save time and refocus their attention on teaching and addressing their learners’ needs.
After uploading data to Azure AI Foundry and creating an index on your data using the integration with Azure AI Search, you can implement the RAG pattern with Prompt Flow to build a generative AI application.
Prompt Flow is a development framework for defining flows that orchestrate interactions with an LLM.
A flow begins with one or more inputs, usually a question or prompt entered by a user, and in the case of iterative conversations the chat history to this point.
The flow is then defined as a series of connected tools, each of which performs a specific operation on the inputs and other environmental variables. There are multiple types of tool that you can include in a prompt flow to perform tasks such as:
Running custom Python code
Looking up data values in an index
Creating prompt variants – enabling you to define multiple versions of a prompt for a large language model (LLM), varying system messages or prompt wording, and compare and evaluate the results from each variant.
Submitting a prompt to an LLM to generate results.
Finally, the flow has one or more outputs, typically to return the generated results from an LLM.
Using the RAG pattern in a prompt flow
The key to using the RAG pattern in a prompt flow is to use an Index Lookup tool to retrieve data from an index so that subsequent tools in the flow can use the results to augment the prompt used to generate output from an LLM.
Use a sample to create a chat flow
Prompt flow provides various samples you can use as a starting point to create an application. When you want to combine RAG and a language model in your application, you can clone the Multi-round Q&A on your data sample.
The sample contains the necessary elements to include RAG and a language model:
Append the history to the chat input to define a prompt in the form of a contextualized form of a question.
Look up relevant information from your data using your search index.
Generate the prompt context by using the retrieved data from the index to augment the question.
Create prompt variants by adding a system message and structuring the chat history.
Submit the prompt to a language model that generates a natural language response.
Let’s explore each of these elements in more detail.
Modify query with history
The first step in the flow is a Large Language Model (LLM) node that takes the chat history and the user’s last question and generates a new question that includes all necessary information. By doing so, you generate more succinct input that is processed by the rest of the flow.
Look up relevant information
Next, you use the Index Lookup tool to query the search index you created with the integrated Azure AI Search feature and find the relevant information from your data source.
The output of the Index Lookup tool is the retrieved context you want to use when generating a response to the user. You want to use the output in a prompt that is sent to a language model, which means you want to parse the output into a more suitable format.
The output of the Index Lookup tool can include the top n results (depending on the parameters you set). When you generate the prompt context, you can use a Python node to iterate over the retrieved documents from your data source and combine their contents and sources into one document string. The string will be used in the prompt you send to the language model in the next step of the flow.
Define prompt variants
When you construct the prompt you want to send to your language model, you can use variants to represent different prompt contents.
When including RAG in your chat flow, your goal is to ground the chatbot’s responses. Next to retrieving relevant context from your data source, you can also influence the groundedness of the chatbot’s response by instructing it to use the context and aim to be factual.
With the prompt variants, you can provide varying system messages in the prompt to explore which content provides the most groundedness.
Chat with context
Finally, you use an LLM node to send the prompt to a language model to generate a response using the relevant context retrieved from your data source. The response from this node is also the output of the entire flow.
After configuring the sample chat flow to use your indexed data and the language model of your choosing, you can deploy the flow and integrate it with an application to offer users an agentic experience.
When you’ve created an Azure AI Search index for your contextual data, you can use it with an OpenAI model. To ground prompts with data from your index, the Azure OpenAI SDK supports extending the request with connection details for the index.
The following Python code example shows how to implement this pattern.
PythonCopy
from openai import AzureOpenAI
# Get an Azure OpenAI chat client
chat_client = AzureOpenAI(
api_version = "2024-12-01-preview",
azure_endpoint = open_ai_endpoint,
api_key = open_ai_key
)
# Initialize prompt with system message
prompt = [
{"role": "system", "content": "You are a helpful AI assistant."}
]
# Add a user input message to the prompt
input_text = input("Enter a question: ")
prompt.append({"role": "user", "content": input_text})
# Additional parameters to apply RAG pattern using the AI Search index
rag_params = {
"data_sources": [
{
"type": "azure_search",
"parameters": {
"endpoint": search_url,
"index_name": "index_name",
"authentication": {
"type": "api_key",
"key": search_key,
}
}
}
],
}
# Submit the prompt with the index information
response = chat_client.chat.completions.create(
model="<model_deployment_name>",
messages=prompt,
extra_body=rag_params
)
# Print the contextualized response
completion = response.choices[0].message.content
print(completion)
In this example, the search against the index is keyword-based – in other words, the query consists of the text in the user prompt, which is matched to text in the indexed documents. When using an index that supports it, an alternative approach is to use a vector-based query in which the index and the query use numeric vectors to represent text tokens. Searching with vectors enables matching based on semantic similarity as well as literal text matches.
To use a vector-based query, you can modify the specification of the Azure AI Search data source details to include an embedding model; which is then used to vectorize the query text.
When you want to create an agent that uses your own data to generate accurate answers, you need to be able to search your data efficiently. When you build an agent with the Azure AI Foundry, you can use the integration with Azure AI Search to retrieve the relevant context in your chat flow.
Azure AI Search is a retriever that you can include when building a language model application with prompt flow. Azure AI Search allows you to bring your own data, index your data, and query the index to retrieve any information you need.
Using a vector index
While a text-based index will improve search efficiency, you can usually achieve a better data retrieval solution by using a vector-based index that contains embeddings that represent the text tokens in your data source.
An embedding is a special format of data representation that a search engine can use to easily find the relevant information. More specifically, an embedding is a vector of floating-point numbers.
For example, imagine you have two documents with the following contents:
“The children played joyfully in the park.”
“Kids happily ran around the playground.”
These two documents contain texts that are semantically related, even though different words are used. By creating vector embeddings for the text in the documents, the relation between the words in the text can be mathematically calculated.
Imagine the keywords being extracted from the document and plotted as a vector in a multidimensional space:
The distance between vectors can be calculated by measuring the cosine of the angle between two vectors, also known as the cosine similarity. In other words, the cosine similarity computes the semantic similarity between documents and a query.
By representing words and their meanings with vectors, you can extract relevant context from your data source even when your data is stored in different formats (text or image) and languages.
When you want to be able to use vector search to search your data, you need to create embeddings when creating your search index. To create embeddings for your search index, you can use an Azure OpenAI embedding model available in Azure AI Foundry.
In Azure AI Search, a search index describes how your content is organized to make it searchable. Imagine a library containing many books. You want to be able to search through the library and retrieve the relevant book easily and efficiently. To make the library searchable, you create a catalog that contains any relevant data about books to make any book easy to find. A library’s catalog serves as the search index.
Though there are different approaches to creating an index, the integration of Azure AI Search in Azure AI Foundry makes it easy for you to create an index that is suitable for language models. You can add your data to Azure AI Foundry, after which you can use Azure AI Search to create an index in the Azure AI Foundry portal using an embedding model. The index asset is stored in Azure AI Search and queried by Azure AI Foundry when used in a chat flow.
How you configure your search index depends on the data you have and the context you want your language model to use. For example, keyword search enables you to retrieve information that exactly matches the search query. Semantic search already takes it one step further by retrieving information that matches the meaning of the query instead of the exact keyword, using semantic models. Currently, the most advanced technique is vector search, which creates embeddings to represent your data.
There are several ways that information can be queried in an index:
Keyword search: Identifies relevant documents or passages based on specific keywords or terms provided as input.
Semantic search: Retrieves documents or passages by understanding the meaning of the query and matching it with semantically related content rather than relying solely on exact keyword matches.
Vector search: Uses mathematical representations of text (vectors) to find similar documents or passages based on their semantic meaning or context.
Hybrid search: Combines any or all of the other search techniques. Queries are executed in parallel and are returned in a unified result set.
When you create a search index in Azure AI Foundry, you’re guided to configuring an index that is most suitable to use in combination with a language model. When your search results are used in a generative AI application, hybrid search gives the most accurate results.
Hybrid search is a combination of keyword (and full text), and vector search, to which semantic ranking is optionally added. When you create an index that is compatible with hybrid search, the retrieved information is precise when exact matches are available (using keywords), and still relevant when only conceptually similar information can be found (using vector search).
Language models excel in generating engaging text, and are ideal as the base for agents. Agents provide users with an intuitive chat-based application to receive assistance in their work. When designing an agent for a specific use case, you want to ensure your language model is grounded and uses factual information that is relevant to what the user needs.
Though language models are trained on a vast amount of data, they may not have access to the knowledge you want to make available to your users. To ensure that an agent is grounded on specific data to provide accurate and domain-specific responses, you can use Retrieval Augmented Generation (RAG).
Understanding RAG
RAG is a technique that you can use to ground a language model. In other words, it’s a process for retrieving information that is relevant to the user’s initial prompt. In general terms, the RAG pattern incorporates the following steps:
Retrieve grounding data based on the initial user-entered prompt.
Augment the prompt with grounding data.
Use a language model to generate a grounded response.
By retrieving context from a specified data source, you ensure that the language model uses relevant information when responding, instead of relying on its training data.
Using RAG is a powerful and easy-to-use technique for many cases in which you want to ground your language model and improve the factual accuracy of your generative AI app’s responses.
Adding grounding data to an Azure AI project
You can use Azure AI Foundry to build a custom age that uses your own data to ground prompts. Azure AI Foundry supports a range of data connections that you can use to add data to a project, including:
Azure Blob Storage
Azure Data Lake Storage Gen2
Microsoft OneLake
You can also upload files or folders to the storage used by your AI Foundry project.
