Registration

I locked 1000 architects in a room until they all agreed on the best solution

What happens when you put 1,000 of the sharpest minds in software architecture together and challenge them to solve a problem? Chaos? Genius? Maybe both. This session will cover the wild process that led to one solution to rule them all. But here's the kicker: the secret weapon wasn’t endless debate or sheer luck—it was something smarter, faster, and way cooler than we could’ve imagined. Come see how we broke the rules and redefined what’s possible in modern architecture. Will you agree with their choice? You'll have to come to this session to find out.

Fast at Fabric - Getting Started

Microsoft Fabric offers a unified analytics platform, but navigating the initial setup can be daunting. This session takes you from zero to fully operational in just 30 minutes. We’ll guide attendees through: - Signing up for a Microsoft Fabric trial and understanding the key requirements. - Common access errors when getting assigned to a tenant and how to resolve them. - Critical tenant admin settings that enable smooth onboarding. - Configuring workspace settings to optimise collaboration. - Spinning up a Fabric capacity resource to start building immediately. Attendees will leave with the confidence to set up their Fabric environment efficiently, ensuring they are fast at Fabric from the very start.

I want to leverage AI – but I’m a little overwhelmed by my options

Whether you’re a beginner in IT or a seasoned professional in another area of IT, you’re likely aware of AI tools like Chat GPT and Copilot. But after supplying a few prompts to one of the LLMs on the market, you’d like to take your skills to the next level – but where should you start? This session will guide you through the maze of Copilot, Copilot Studio and AI Foundry so that you can start making sense of everything AI on the Microsoft Platform – which encompasses many non-Microsoft LLMs. This short session should help to cut through the noise and leave you with some options of the next steps on your AI journey.

How I Reduced My Power BI Dataset By 60%

How to optimize Power BI Datasets to ensure that they can run as fast as possible with the smallest amount of memory possible. The session will cover the following topics: Data Modelling with the Star Schema Looking at columns in the dataset Data Types and how this affects the size of your dataset Using DAX Studio for analysis of datasets Real world optimizations that I put into practice

Local Intelligence: Running LLMs Natively, No Cloud Required

Imagine an AI that runs entirely offline—fast, private, and directly connected to your application’s logic. In this session, we’ll explore how locally hosted large language models (LLMs) can interpret natural language prompts and invoke real-time functionality without ever touching the cloud. From smart environments to embedded devices, we’ll demonstrate how this approach unlocks new possibilities for responsiveness, autonomy, and data sovereignty. We’ll wrap by showing how .NET 9’s new AI-native method invocation makes this not just possible—but practical.

Short Break

Considerations for an Enterprise Power BI or Fabric Implementation

When implementing Power BI or Microsoft Fabric in any enterprise, there are many options that the project team faces, and the choices are not always obvious. The different approaches are often dictated by organisational constraints, which are often different for many organisations. In this session, Craig will share with the audience the main areas of consideration which impact how your company goes about implementing Power BI or Microsoft Fabric in a secure and governed fashion. The audience will walk away from the session with a basic understanding of what their organisation needs to consider for the effective management of their analytics, reporting, and AI environments.

LLMs + Semantic Models = Smarter SQL

Getting accurate Text-to-SQL results isn’t magic, you need two things: smarter language models and a clear semantic model that explains what your data actually means. LLMs are getting better at understanding natural language, but without structure, they still tend to guess or hallucinate. To explore this, I built a small experiment using a local DB setup, a semantic layer over a typical dataset , and a set of benchmark questions. I tested different models like o3-mini , Qwen3 to see how well they generate SQL when given proper context. The big takeaway? With the right metadata, even smaller local models can do surprisingly well, and running accurate Text-to-SQL on everyday hardware isn’t just possible, it’s only a matter of time.

Real-Time Analytics

Vision AI

Social Event