How AI Copilots can empower IT professionals

Master Cloud Cost, Cloud Service Policies and more — with ease

Artificial intelligence (AI) is not a futuristic fantasy, but a practical reality that can help IT professionals with various tasks and challenges in the cloud era. In this article, we will explore how AI assistants can empower IT professionals to master cloud costs and policies, using some examples from Microsoft and OpenAI.

Copilot: Your personal guide to cloud cost optimization

One of the challenges of cloud computing is to optimize the cost and performance of cloud resources. Microsoft Cost Management is a service that helps customers monitor, analyze, and optimize their cloud spending across Azure and other providers.

But what if you could have a personal guide that could answer your questions, provide insights, and give recommendations for cost optimization? That’s where Copilot comes in.

Copilot is an AI assistant that uses natural language processing (NLP) and machine learning (ML) to understand your queries and goals, and to generate relevant and actionable answers. Copilot is like a friendly expert that can help you with your cloud cost management.

For example, you can ask Copilot questions like:

• How much did I spend on Azure last month?
• Which resource group has the highest cost?
• How can I reduce my cost by 10%?

Copilot will then analyze your data and provide answers in natural language, along with charts and tables to visualize the data. Copilot will also suggest best practices and tips for cost optimization, such as resizing or deleting unused resources, applying reserved instances or spot instances, or using Azure Advisor recommendations.

Copilot is currently in preview and available for free for all Microsoft Cost Management customers. You can access Copilot from the Microsoft Cost Management portal or from Microsoft Teams.

OpenAI Codex: Your creative partner for API management policies

Another challenge of cloud computing is to define and enforce policies for API management. API management is a process that involves creating, publishing, securing, monitoring, and analyzing APIs that connect applications and data across cloud and on-premises environments.

API management policies are rules that define the behavior and functionality of APIs, such as authentication, authorization, caching, throttling, transformation, logging, etc. Writing API management policies can be tedious and error-prone, especially for complex scenarios or multiple APIs.

But what if you could have a creative partner that could generate code from natural language descriptions or examples? That’s where OpenAI Codex comes in.

OpenAI Codex is an AI system that can generate code from natural language descriptions or examples. OpenAI Codex is powered by GPT-3, a large-scale language model that can generate coherent and diverse texts on various topics.

One of the applications of OpenAI Codex is to generate API management policies from natural language prompts or examples. For example, you can provide a prompt like:

• Generate an API management policy that validates the JWT token in the Authorization header and extracts the user ID from the token

OpenAI Codex will then generate a valid XML code for the policy, such as:

<policies>
  <inbound>
    <validate-jwt header-name="Authorization" failed-validation-httpcode="401" failed-validation-error-message="Unauthorized">
      <openid-config url="https://example.com/.well-known/openid-configuration" />
      <required-claims>
        <claim name="sub" match="any" />
      </required-claims>
      <output-token-variable-name>jwt</output-token-variable-name>
    </validate-jwt>
    <set-variable name="userId" value="@((string)((JwtOutput)context.Variables["jwt"]).Claims["sub"].Value)" />
  </inbound>
</policies>

This way, OpenAI Codex can help you write API management policies faster and easier, without having to remember the syntax or look up the documentation.

Vectors and Embedding Stores: Your secret weapons for semantic search

A third challenge of cloud computing is to find relevant information from large collections of data. Whether you are looking for documents, images, videos, or products, you want to be able to search by meaning rather than by keywords.

But how can you represent meaning in a way that computers can understand? That’s where vectors and embedding stores come in.

Vectors are numerical representations of data that capture their semantic properties. For example, you can represent words as vectors based on their usage in texts. This way, words that have similar meanings will have similar vectors.

Embedding stores are databases that store vectors along with their metadata. For example, you can store word vectors along with their corresponding words. This way, you can query the embedding store by words or by vectors.

One of the applications of vectors and embedding stores is to perform semantic search. Semantic search is a type of search that uses vectors to find data that are semantically related to the query. For example, you can use semantic search to find products that are similar to a given product, or images that are similar to a given image.

Semantic search can be implemented using OpenAI’s DALL-E and CLIP models. DALL-E is a model that can generate images from natural language descriptions. CLIP is a model that can learn from any kind of image and text pair. Together, they can create and search for images based on their meaning.

For example, you can use DALL-E and CLIP to generate and search for images of “a dragon made of flowers”. DALL-E will generate several images of dragons made of flowers, and CLIP will rank them based on their relevance to the query. You can also use CLIP to search for existing images that match the query.

This way, vectors and embedding stores can help you find relevant information from large collections of data, based on their meaning rather than their keywords.

Vectors and embedding stores are currently available for developers who sign up for the OpenAI Beta. You can access vectors and embedding stores from various platforms and tools, such as OpenAI Playground, OpenAI API, or OpenAI Jupyter Notebooks.

The Bottom Line

AI assistants are becoming more capable and accessible for IT professionals who want to master cloud costs and policies. By using AI assistants like Copilot, OpenAI Codex, and vectors and embedding stores, IT professionals can save time and effort, improve accuracy and quality, and enhance their productivity and creativity. AI assistants are not meant to replace human intelligence or expertise, but rather to augment and complement them. AI assistants are here to empower IT professionals to achieve their goals and solve their problems in the cloud era.