Build Agents Once, Run Them Anywhere: AI Systems That Survive Change
-
Isha Salania
Microsoft
Cloud & AI Solution Engineer
3 Ways to Rebuild the Data Stack for Agents
July 9, 2026
14:10
–
14:40
·
30
min
Stage 12
What this session covers
AI agents are becoming core users of the data tools we’ve used for decades. Seemingly everyone has a dozen agents querying their data warehouse, building dashboards, or writing their data pipelines.
This session will share what we’ve learned at MotherDuck while building data products for users that rely on AI agents. We’ll cover:
- Tradeoffs of owning the agent harness vs. the BYO agent model
- The role of CLIs and Model Context Protocol in the data stack
- How to create products agents can discover, sign up for, and use
World Congress 2026 Europe
Related talks
Your Data, Your Agents: Building a Multi-Agent Workflow with Microsoft Fabric
-
Hanna Schwab
teccle group
Data and AI Lead
5 things I wish I hadn’t done building my AI agent
-
Shachar Azriel
Baz
VP Product
When Agents Meet Legacy: Never Change a Running System
-
Michael Friedrich
GitLab
Principal Developer Advocate