Victoria Tech Week Session Hosted by AOT Technologies

As part of 2026 Victoria Tech Week, AOT Technologies' Data and AI practice brings three live sessions to the room. Marie Anik Paradis shows non-technical professionals how to teach an AI to do a task they repeat every week, no code. Shuo Xu takes apart why accurate data still takes hours to become a decision. Abin Antony covers what breaks when an LLM is talked into something, and the defences that hold. Free pizza, drinks, and refreshments will be provided.

Info: This is a live, hands-on session about teaching AI to do the work you already know how to do. We’ll start by showing how AI created the session’s slides and handout from a playbook written entirely in plain English. You’ll see that process happen live in the first five minutes.

Then we’ll break down how the playbook works and how to turn a repeatable task into an AI skill, without coding, complex automation, or “magic” prompts. The audience will choose a real task, and we’ll build a working AI skill for it together, on stage.

By the end, you’ll have seen the full process from task to working AI skill, along with a practical template and recipe you can take back to work and try yourself.

Info: We’ll follow a real-world failure from the moment it begins to the moment someone finally acts, and unpack why three hours can disappear even when the data pipeline and dashboard are working perfectly.

Live, we’ll break the delay into five distinct latencies: data, analytical, context, decision, and action. We’ll trace where each occurs, what causes it, and how delays compound across systems, teams, and decisions.

The talk will then connect each latency to the capabilities of a modern data platform, showing which problems technology can actually solve, and which require changes in context, workflows, or decision-making. The goal is to leave you with a practical framework for diagnosing why “real-time data” still produces slow decisions, and where to focus your next investment.

Info: This talk takes a practical look at what happens when an LLM cannot reliably distinguish between trusted instructions and malicious ones. We’ll show how prompt injection attacks work, how seemingly harmless inputs can redirect an AI system, and why traditional filters and guardrails aren't a complete solution.

We’ll then examine the defences that actually help: designing safer AI workflows, limiting what models can access or execute, validating inputs and outputs, and using security gateways to create additional control points around LLMs.

Rather than focusing on how to make a model impossible to fool, the session asks a more useful question: what happens when it is? You’ll leave with a clearer understanding of the attack surface, the practical controls that reduce risk, and how to build AI systems that remain safe even when the model gets something wrong.

Event venue & nearby stays

AOT Technologies, 325-1207 Douglas Street, Victoria, Canada

Concerts, fests, parties, meetups — all the happenings, one place.

Get the app →

Discover more events by tag