Data Platform and Cloud - November User Group Meeting

Hosted at: Ithaca Energy offices, Aberdeen

Event Sponsorship: Kubrick

***

We’re excited to have you join us for an evening of learning, networking, and good company! Here’s what we have planned:
18:00 - Doors Open

• Grab some pizza, meet fellow members, and get settled in!

18:35 - Welcome & Introductions

• A quick hello to kick things off and set the tone for the evening.

18:40 - Lightning Talk (20 minutes)

• Speaker : Geoff Jones
• Topic: Your AI Agent Doesn’t Know What a Barrel Is
  Every organisation is now building AI agents, and most of them fail in the same way. The agent can see tables and columns, but it has no idea what the business means by them. Ask it “what’s our revenue per active customer this quarter?” and it has to guess. Gross, net or recognised revenue? Is an active customer someone who logged in, or someone who paid? Does the fiscal year start in January or February? Which of the 47 revenue tables is the one Finance actually trusts? The agent picks something and gives you a confident answer that’s wrong. Then it does all that guessing again from scratch the next time someone asks.
  This talk looks at that context problem and at one practical way to solve it: Databricks Genie Ontology. It’s an enterprise context layer that sits between your data and any agent. It combines two kinds of knowledge:
  Modelled knowledge: governed KPI definitions, business glossaries and domains that people curate and certify.
  Learned knowledge: definitions, rules and sources of truth picked up automatically from the queries, notebooks, dashboards and pipelines your teams already use.
  A ranking algorithm called OntoRank then weighs that knowledge by authority and relevance and respects each user’s permissions. Any agent can use the context over MCP, so you’re not tied to one tool.
  We’ll walk through how it works step by step, show worked examples, and look at evaluation results on real-world questions. You’ll leave with a clear picture of where human curation pays off, where automated learning does the work for you, and how to start building a context layer for your own data.

19:15 - Main Presentation (40 minutes)

• Speaker: Steven Wallace
• Topic: AI Agents: From Perception to Autonomous Action
  Recent advances in foundation models have given rise to AI agents capable of reasoning, planning, and acting autonomously across complex tasks. This talk explores the shift from static, single-purpose deep learning models towards agentic systems that combine perception, multi-modal understanding, and decision-making to operate independently in real-world environments. Drawing on examples from natural language processing and beyond, we examine the opportunities of building AI agents that are capable of carrying out complex multi-step tasks.

19:55 - Closing Remarks & Thank You

• A chance to wrap things up, share feedback, and express our appreciation to everyone.

20:00 - Event Ends

• We’ll bring the evening to a close, but feel free to keep the conversation going!

We look forward to seeing you there!

***