Microsoft AI Summit Key Takeaways
Microsoft 365 Copilot:
- How AI Is Empowering Human Achievement: By revolutionising productivity.
- Record-Breaking Success: Microsoft 365 Copilot is the fastest-growing cloud service with the most successful product launch.
- Core Features: It enhances productivity with immersive, accessible features and measurable ROI.
- The live demonstration: Included capabilities of email summarisation and prioritisation, Excel integration (to create charts and tables from chat commands), Teams alerts and opportunities, and customisation options, including memory additions for specific tasks.
Copilot Studio:
- Create Agents with natural language to define workflows and tasks
- Key Functions: Draft emails and team messages, utilise knowledge base articles, triggered based on first or third-party actions along with self-improvement capabilities.
GitHub Copilot:
- New extensions and enhancements: Review multiple files simultaneously and infuse context from external applications, noting there were some challenges in the live demo.
Azure AI Foundry:
- Unified developer interface: With a dual-interface platform to benefit both technical and non-technical users, enabling the creation of microservices and LLM application testing with a demonstrated real-time voice API.
Microsoft Fabric:
- SaaS platform:
- Has 150+ connectors to other data platforms.
- Features lineage views in Lakehouse to track data usage and Snowflake integration as a storage layer.
- Uses natural language processing with Copilot to generate code in notebooks and create Power BI reports.
Microsoft Security Unified Platform
- Incident management: Summarises information with timelines, Identifies attack patterns, suggests actionable responses, and generates incident reports.
Live Interviews:
Matt Comyn (CBA CEO)
- Focus on powering customer experiences.
- Importance of foundational elements like data strategy.
- Leveraging capabilities and reusable assets.
Leah Banter (CEO, Wesfarmers)
- Acknowledged late entry to digital transformation (for example, Bunnings).
- Focused on enhancing customer experience.
- Early adopters in three divisions with over 1,000 licenses testing.
- Significant feedback from Finance and HR showing 50% productivity gains.
Virginia Briggs (CEO, MinterEllison)
- Early experimentation with AI since 2016.
- Piloting Microsoft Copilot with 770 licenses.
- 89% report saving 2-5 hours per week.
- 71% report increased productivity.
Matt Comyn (CBA)
- Rolled out 20,000 licenses for GitHub and M365 Copilot.
- Highlighted 20% improvement in match rates (specific details unclear).
To learn how we can assist you in taking your next step in adopting AI to improve your productivity, please reach out to our team for a confidential conversation HERE.
Breakout Sessions
Session 1: Simplifying LLM Development with Prompty
- Tool Highlight: Prompty simplifies LLM app development.
- Key Features:
- Seamless integration with Visual Studio Code and GitHub.
- Learn to build multi-model LLM agent architecture, evaluation, deployment, and monitoring.
- More Info: Prompty.ai
Session 2: Building RAG Applications with Azure AI Search
-
Focus Areas:
- Basics of vector embeddings and search strategies.
- Full-text, vector, and hybrid search with Azure AI Search.
- Data ingestion processes: manual and integrated vectorization.
- Evaluation strategies using AI Studio and Promptflow-evals SDK.
- Efficient vector storage with quantization.
- Security best practices: private network deployment, user authentication, and data access control.
- Learning Outcomes:
-
- Understanding vector search strategies and Azure AI Search options.
- Streamlining data ingestion and evaluation for high-quality RAG app answers.
- Ensuring robust security and efficiency in production-ready applications.