Google just proved it live.
In the latest video they asked one question inside their own IDE: "Calculate our monthly average order value between these two dates." Full workflow, real BigQuery access, real answer. Then they opened Claude Code, installed one plugin, asked the exact same question — and got the exact same workflow. Same MCP calls, same tables searched, same number back.
I mentor data analysts on how to actually work with AI in analytics, and this is the exact thing most of them get wrong first: they pick a tool like it's a personality test, then wonder why it "can't see" their data. The tool was never the bottleneck. The setup was.
Here's the part that matters: the skill and the connector are the portable unit, not the interface. Install the Data Agent Kit plugin — one command, pulled from GitHub or the Anthropic marketplace — and Codex, Claude Code, and Gemini CLI all get the same secure MCP access to BigQuery, Cloud SQL, AlloyDB, Spanner, and the same "developing with BigQuery" skill that keeps the SQL syntax correct. Nothing is native-only. Nothing is the "lite" version in your CLI of choice.
They even ran the follow-up — "January looks lower than prior months, break it down by order type" — in the second tool and got the identical root cause: a new B2B wholesale segment. Not a similar answer. The same answer, from a different terminal.
If your AI coding tool can't reach your actual data stack yet, that's not a limitation of the tool. That's a setup you haven't done.
Want a stronger data analyst role or a raise? Grab the FREE Product Analyst Playbook and get the exact roadmap to your next offer.
