### 🗓️ Event Details
Join the Snowflake TLV Users Group for an evening focused on real-world AI agents and the next generation of AI tooling on Snowflake. Hear how WSC Sports built production-grade Cortex Agents workflows- from YAML-defined semantic models to governed AI agents; and explore the latest capabilities around CoCo, Claude, and Agent Skills with Snowflake.
# Agenda
## 18:00 – 18:30 | Gathering & Networking
## 18:30 – 18:55
### LLM and Data Architecture for Decision Making
#### Amit Pollak, BI Team Leader, WSC Sports
What does it take to run AI agents in production at scale?
In this session, we’ll walk through how the BI team at WSC Sports built an operational platform for Snowflake Cortex Agents- enabling anyone in the organization to ask natural-language questions against governed data models and get instant, accurate answers without writing a single line of SQL.
Our guiding concept: “From YAML to Agent” — a user defines a semantic model in a simple YAML file, submits a PR, and the platform takes it all the way to a live, governed, monitored agent running in Snowflake.
A practical look at what production-grade AI on top of a data platform actually looks like.
## 18:55 – 19:15
### From Launch to Trust: Building and Maintaining Reliable Data Agents
Haim Moscovici, Data Analytics Tech Lead, WSC Sports
Building a data agent is one thing. Getting users to trust it enough to actually use it? That’s the real challenge.
The 80/20 rule doesn’t work for agents answering business questions — you need to aim for excellence.
In this session, we’ll explore the reliability standard that actually matters, and how to achieve it through a phased rollout approach focused on validation, monitoring, and continuous improvement.
## 19:15 – 19:30 | Break & Networking
## 19:30 – 20:10
### When CoCo Meets Claude: Next-Gen AI for Your Snowflake Data
Matan Sarig, Senior Solution Engineer, Snowflake
What happens when you combine Snowflake’s native Cortex Code (CoCo) with the deep reasoning of Anthropic Claude? You get data agents that actually know what they’re doing.
We’ll look at the best ways to run CoCo — whether you're in Snowsight, the CLI, your Desktop IDE, or using a direct Claude integration. We’ll check out the latest Agent Skills to see how easy it is to give your assistant custom superpowers, and demo real-world tasks where the Claude + CoCo combo brings significantly more value than standard AI tools.
## 20:10 – 20:30 | Mingling & Networking
