DataKitchen Webinar
Stop Clicking, Start Asking: The AI Playbook for Data Quality & Observability
June 7th, 2026; 12 pm EST / 4 pm GMT
Data quality work has always been a clicking job, and AI is about to change that. Today, you open the dashboard, scan the tiles, drill in, and file the failure for someone else. The dashboard (or incident list) is fast once you know it, but it makes you do all the thinking in your head while the tool sits silent.
Here is what is new. MCP (Model Context Protocol) lets an LLM like Claude or ChatGPT talk directly about your data quality, in plain English, with real data and real history. DataKitchen's open source TestGen now supports it. The protocol is here, it works, and you can download it today.
That opens up two new ways of working.
-
AI Dialogue. You ask the tool questions in plain English. Why did this fail? Is it new? Show me the rows. What changed since last Tuesday? What test is missing from this table? The tool answers with the actual data and the actual history. The investigation that used to span five screens fits in one window.
-
AI Delegation. AI agents are autonomous programs that take actions on your behalf. Can you safely let one change your data? What guardrails do you need? Where does delegation actually pay off, and where will it burn you?
This webinar is the playbook for the full arc. Where the dashboard still wins. What AI dialogue unlocks. When is it safe to turn on AI Agents?
You will leave with:
- A diagnostic conversation pattern you can reuse for any failure
- A three-lane delegation framework for what is safe to automate today
- A clear sense of which mode to reach for at any given moment
Who this is for: Data engineers, data quality leads, observability owners, and anyone running a test suite where the morning triage has become the worst hour of the week.