Microsoft’s AI Momentum: Copilot Evolves, GPT-4.1 Launches, and Llama 4 Joins Azure

As Microsoft celebrates its 50th anniversary, the company has unveiled a series of AI advancements that underscore its commitment to innovation. From enhancing Copilot’s capabilities to introducing new AI models in Azure, these developments are set to redefine user experiences and enterprise solutions.

Copilot Becomes Your Personalised AI Companion

Microsoft has transformed Copilot into a more intuitive and proactive assistant. Key enhancements include:

Personalized Memory: Copilot now remembers user preferences and past interactions, allowing for tailored assistance. Users have full control over what Copilot retains, ensuring privacy and customization.
Proactive Actions: With the new “Actions” feature, Copilot can autonomously perform multi-step tasks, such as booking travel or making reservations, through integrations with services like Expedia, TripAdvisor, and OpenTable.
Enhanced Integration: Copilot’s integration extends across Windows 11, iOS, and Android platforms, offering features like visual awareness through smartphone cameras and the ability to monitor on-screen activity to provide context-aware assistance.
New Tools: Introduction of “Pages” and “Deep Research” tools allows users to organize projects and conduct in-depth research more efficiently.
Generative Podcasts: Copilot can now create personalized podcasts based on user interests and previous interactions, offering a new medium for information consumption.
Improved Bing Search: Integration with Bing Search now includes generative AI responses within standard search results, enhancing the search experience with deeper insights.

These updates aim to make Copilot a central AI companion that seamlessly assists users across various aspects of their digital lives.

Read the full details here.

Copilot Studio Introduces ‘Computer Use’ for UI Automation

Microsoft has introduced a new feature called “computer use” in Copilot Studio, now available in early access. This capability allows AI agents to interact directly with websites and desktop applications by simulating human actions such as clicking buttons, selecting menus, and typing into fields on the screen. This means agents can automate tasks even when no API is available, effectively treating any graphical user interface as a tool.

The “computer use” feature enhances automation by enabling agents to adapt to changes in applications and websites in real time, using built-in reasoning to handle issues autonomously. This ensures continuous operation without manual intervention. Additionally, it is built on Copilot Studio’s security and governance frameworks to maintain organizational compliance.

Read the full article here.

GPT-4.1 Model Series Debuts on Azure AI Foundry

Microsoft has announced the availability of the GPT-4.1 model series, comprising GPT-4.1, GPT-4.1 Mini, and GPT-4.1 Nano, on the Azure OpenAI Service and GitHub. These models offer significant enhancements in coding, instruction following, and long-context processing, making them valuable tools for developers.

Key Features:

Advanced Coding and Instruction Following: GPT-4.1 is optimized for handling complex technical and coding tasks. It generates cleaner front-end code, accurately identifies necessary changes in existing code, and consistently produces outputs that compile and run successfully.
Extended Context Window: All three models support inputs of up to one million tokens, allowing for the processing and understanding of extensive context in a single interaction. This is particularly beneficial for tasks requiring detailed and nuanced understanding, as well as multi-step agents that increase context as they operate.
Improved Instruction Following: The models excel at following detailed instructions, especially in scenarios involving multiple requests. They are more intuitive and collaborative, enhancing their applicability across various applications.

Fine-Tuning Capabilities:

Microsoft plans to enable supervised fine-tuning for GPT-4.1 and GPT-4.1 Mini later this week. This will allow developers to customize these models to their specific business needs, aligning responses with organizational tone, domain terminology, and task workflows.

Find out more here.

Llama 4 Models Now Available in Azure AI Foundry

Microsoft has introduced Meta’s Llama 4 models, Scout and Maverick, to Azure AI Foundry and Azure Databricks, enhancing the development of advanced, multimodal AI applications.

Model Highlights:

Scout: A 17B parameter model supporting up to 10 million tokens, optimized for tasks like multi-document summarization and complex reasoning over extensive datasets.
Maverick: A 17B active parameter model with 128 experts (400B total), offering multilingual support across 12 languages and optimized for chat and image understanding.

Architectural Innovations:

Early Fusion Multimodality: Llama 4 models process text, images, and videos as a unified sequence of tokens, enabling seamless multimodal understanding and generation.
Sparse Mixture of Experts (MoE): This architecture activates only a subset of expert models per input, enhancing efficiency and scalability without compromising performance.

These models are now available through Azure AI Foundry, providing developers with tools to build and deploy customized generative AI solutions.

Read the full update here.

Microsoft’s latest AI advancements reflect a strategic push to integrate intelligent, personalized, and proactive capabilities across its platforms. From empowering users with a more capable Copilot to providing developers with cutting-edge models like GPT-4.1 and Llama 4, Microsoft is setting the stage for a new era of AI-driven experiences.

For more information or assistance with integrating these AI solutions into your business, feel free to contact the 365 Mechanix team.