.png)
.png)
.png)
Agentic AI for Finance & Analytics Teams
Practical applications, not just the hype
AI agents are about to reshape how finance and analytics work gets done. The only question is whether you lead that shift or scramble to catch up. Agentic AI has moved beyond presentation slides to something teams can genuinely use, but the gap between a polished demo and a workflow you'd trust at month-end is where most teams get stuck.
This hands-on, 1-day in-person Masterclass cuts through the hype and shows what Microsoft has actually shipped for finance and analytics work - Copilot in Excel, the Microsoft 365 Finance agent, Fabric data agents over your governed data, and custom agents built in Copilot Studio. You'll see these applied to real scenarios like variance analysis, commentary, forecasting and reconciliations.
You'll see where agents save real hours, where they can't yet be trusted, and how to deploy them safely with the accuracy, auditability and governance finance demands. You'll leave with a practical understanding of the tools and a 90-day plan to put one agent to work on your own team.
What you'll learn?
- What "agentic" actually means, the spectrum from assisted to autonomous, and how to tell production-ready capability from hype
- How to drive Copilot in Excel (edit, plan and chat modes) to build models, analyses and forecasts from plain English and where its hard limits are
- How the Microsoft 365 Finance agent and Fabric data agents extend AI from a single workbook to your governed enterprise data
- How to build and automate your own finance/analytics agent in Copilot Studio, including a real month-end workflow
- How to handle the parts that actually matter in finance, accuracy, auditability, human review and approval, governance and cost
This class will be delivered in-person in Central London, UK, on 17 May 2027 (10:00 am–5:00 pm BST).
This training is available as a single Masterclass or as part of our Premium (In-Person) or Ultimate (In-Person) Global Excel Summit ticket.
What's included
Who is it for
Suitable for individual contributors through to finance and analytics team leads weighing up where agents fit. Typical attendees include:
- Finance and analytics professionals
- FP&A
- Controllers
- Accountants
- Finance business partners
- BI professionals
- Data analysts
- Excel users, who are confident in Excel and want to move from "I've heard about AI agents" to "I can use them, safely, on real work."
Download Course Brochure
Course contents
- The spectrum: assisted, then automated, then agent, goal-driven, tool-using, self-checking
- Where the 2026 Microsoft stack really sits: Copilot, the Finance agent, Copilot Studio, Fabric data agents, Cowork
- An honest map of what's production-ready versus preview versus marketing
- The core misconception: an agent is not a chatbot, and autonomy is a dial, not a switch
- Editing with Copilot, edit, plan and chat modes, and choosing your model
- Live build: messy finance data into a clean model, then variance analysis, a forecast (with Python in Excel) and a chart, all from natural language
- Using plan mode for reviewable, multi-step work
- A deliberate, teachable error and how to catch what the agent gets wrong
- The hard limits: current workbook only, no access to enterprise data, direct edits
- The role-based finance experience: Finance Home, Finance Chat and conversational access to ERP data
- Working in context across Excel, Outlook and Teams while protecting ERP integrity
- Practical plays: variance commentary, reconciliation assist, close acceleration
- Where it fits versus general-purpose Copilot
- Why finance is the hardest case: a confidently wrong number is worse than none
- Grounding, sources and citations, and spotting hallucination patterns
- Designing review gates, and deciding what an agent may never finalise alone
- Exercise: find the errors in an AI-built model
- Recap and key takeaways
- Why enterprise questions need more than an Excel-bound agent
- OneLake and Fabric data agents: natural-language answers over your Lakehouse, Warehouse and semantic models
- How an Excel and Power BI audience plugs in, via semantic models, no code required
- Demo: questioning a governed finance dataset, with grounding and data lineage
- Creating an agent in plain language, grounding it in your data, and giving it actions
- Agent flows and workflows: automating a real month-end variance pack, pull data, analyse, draft commentary, route for approval
- Connectors, MCP and multi-agent orchestration at a glance
- When a deterministic flow beats generative reasoning, and when it doesn't
- The credit-and-metering reality, why autonomous agents cost more, and why CFOs watch AI spend
- Security and compliance boundaries, data residency and model choice
- Avoiding agent sprawl: who builds, how you monitor, how you evaluate quality
- A practical governance checklist to take away
- Choosing the right first use case: high value, low risk, measurable
- Pilot, measure, scale and change management for finance teams
- Build your own first-agent canvas
- Recap, resources and Q&A
About your trainer
Leon Gordon
Leon Gordon is a Microsoft Data Platform MVP and the founder of Onyx Data. He specialises in the Microsoft Data Platform, Microsoft Fabric, Foundry, Power BI and Azure SQL and works at the intersection of data analytics and agentic AI for global enterprise and finance teams.
Leon designs and delivers engaging, hands-on masterclasses, in person and online, that turn the platform's fast-moving capabilities into skills people can use the next day. His focus is practical and outcomes-driven, not features for their own sake, but what actually works in production.
What people think of
Leon Gordon
Leon is the best public speaker on AI I have heard!
Leon is the best public speaker on AI I have heard and one of the best keynote speakers too. Makes it all very understandable as well, which is all too rare. Genuinely interesting background too.
His contributions undoubtedly enhance the overall experience for participants!
Leon Gordon captivated the audience with insightful and engaging delivery. His presentation style is both dynamic and engaging, making him a standout speaker in the [AI] field. I highly recommend Leon as his contributions undoubtedly enhance the overall experience for participants.
I wholeheartedly recommend Leon as a speaker and am confident that his presentations will add significant value to any event or educational setting.
I have had the opportunity to witness Leon presentations on multiple occasions, and I am consistently impressed by his knowledge, clarity and engaging communication style. He has a deep understanding of complex concepts and can effectively distill them into clear, concise, and accessible explanations. His ability to simplify intricate topics without losing their essence makes his presentations valuable for both experts and novices in the field.
Training requirements
Attendees should already be comfortable with day-to-day Excel use. Please bring:
- A laptop with Microsoft Excel (desktop version recommended) or another application that can open .xlsx files. A phone or tablet alone will not be sufficient.
- A Microsoft account, signed in before the session, if you would like to follow the optional Microsoft 365 exercises.
Optional access: attendees with Microsoft 365 Copilot, Copilot in Excel, Copilot Studio or Microsoft Fabric may follow selected demonstrations hands-on. Availability depends on your organisation's licences and administrator settings, so equivalent exercises are provided for everyone without access.
Buy ticket
Buy 365-Day Digital Masterpass
If you continue experiencing issues, you can send it to us via email to info@globalexcelsummit.com
.gif)
Early Bird registration is open
18–19 May 2027 · London & Online.
Save 20% – limited to 100 in-person and 100 online tickets.
