Summary
Microsoft Build 2024 highlighted key AI advancements across apps and services. Major announcements include a Custom Generative Mode for handling varied invoices, enhanced Azure AI Search with improved accuracy and vector support, and the general availability of Azure AI Studio with advanced tools and models. Key updates to Azure OpenAI Service feature GPT-4o and refined AI interactions. The event also introduced the multimodal Phi-3 model for visual reasoning and AI capabilities in Azure Database for PostgreSQL. Azure Infrastructure updates include the simplified provisioning of VMs with Azure Compute Fleet and new high-performance VM series. Overall, the event underscores Microsoft’s commitment to pushing AI boundaries across various industries.
1. Custom Generative Mode in Preview
Imagine managing invoices from various suppliers, each with different formats. With the Custom Generative model, you can start with a single sample invoice and define fields like “Invoice Number,” “Date,” “Total Amount,” and “Supplier Name.” As you add more invoices, the model learns and adapts, accurately extracting these fields even from varied layouts. If the model misses a field, you can correct it, and it will improve over time, saving you significant time and effort.
Read more here: Introducing Model Customization for Azure AI – Microsoft Community Hub
2. Azure AI Search Enhancements
Azure AI Search now includes several major updates:
- Support for binary vector types in vector search.
- Enhanced response accuracy with new vector and hybrid search capabilities.
- Simplified image processing, vectorization, and search via built-in tools like Microsoft Azure AI Vision and Azure OpenAI Service embedding models.
- Direct data integration from Microsoft Fabric to AI Search with OneLake.
Read more here: Revolutionizing search and retrieval with Azure AI Search (microsoft.com)
3. Azure AI Studio Now Generally Available
Azure AI Studio offers a user-friendly interface and code-first capabilities for developers to explore AI tools, orchestrate multiple APIs and models, and test AI innovations. Key features include:
- Access to the latest foundation models as a service (MaaS), allowing developers to build generative AI apps using Pay-as-you-go inference APIs.
- Comprehensive AI toolchain and app services, including data integration and prompt orchestration.
- Prompt Flow for workflow orchestration of multimodal models, including images as inputs and outputs.
Read more here: Azure AI Studio – Creating and scaling your custom copilots (microsoft.com)
4. Azure OpenAI Service Key Advancements
The Azure OpenAI Service has introduced several updates, transforming AI applications across various business scenarios:
- Availability of GPT-4o in Azure AI Studio.
- Fine-tuning capabilities for GPT-4.
- Assistants API for enhanced AI interactions.
- Preview of messaging insights for WhatsApp through Azure Communication Services.
Read more here: Develop and deploy generative AI apps responsibly with Azure AI Studio (microsoft.com)
5. Microsoft Adds Multimodal Phi-3 Model
The new Phi-3-vision model in the Phi-3 family supports general visual reasoning tasks and responses involving charts, graphs, and tables. It can process images and text inputs and generate text responses.
Read more here: New models added to the Phi-3 family, available on Microsoft Azure | Microsoft Azure Blog
6. AI Capabilities in Azure Database for PostgreSQL
The PostgreSQL extension for Azure AI is now generally available, enabling developers to build rich generative AI apps using large language models (LLMs). Key features include:
- Generating LLM-based vector embeddings for similarity searches.
- Utilizing Azure AI Language for sentiment analysis and language detection.
- Real-time predictions with pre-trained machine learning models on Azure ML.
- Real-time text translation using Azure AI Translator.
Read more here: Build an AI-powered app with PostgreSQL within 10 mins (microsoft.com)
7. Azure Infrastructure Updates
Azure Compute Fleet in Preview
Azure Compute Fleet simplifies provisioning Azure compute across various virtual machine types, availability zones, and pricing models. It allows customers to manage up to 10,000 VMs with a single API call.
Read more here: Announcing the Public Preview of Azure Compute Fleet – Microsoft Community Hub
New Azure Virtual Machine Series
- The Cobalt 100 Arm-based VM, featuring Microsoft’s custom Cobalt 100 processor, offers up to 40% improved performance for general-purpose and cloud-native workloads.
- The ND MI300X v5 VM, optimized for AI and high-performance computing, features the AMD Instinct MI300X AI accelerator and NVIDIA Quantum-2 CX7 InfiniBand, providing significant memory and bandwidth capabilities.
Read more here: Announcing the preview of new Azure VMs based on the Azure Cobalt 100 processor – Microsoft Community Hub
In conclusion, Microsoft Build 2024 is shaping up to be a groundbreaking event with a strong focus on AI. From advanced models and AI-powered database capabilities to enhanced infrastructure and multimodal models, Microsoft is pushing the boundaries of what’s possible with AI. Stay tuned for more updates as these innovations continue to evolve and impact various industries.
Accelerate with Allata
At Allata, we guide our clients towards Digital Excellence. Are you interested in learning more about the impacts of AI? Let’s talk.
