Clear the ad-hoc queue without losing control of the SQL
Marketing analysts field a constant stream of "quick questions" that each need a query. DataLens AI drafts and runs that SQL from a plain-English question, against your schema and definitions, and shows the query, sources and cost so you can review it like a colleague's pull request instead of trusting a black box.
What gets in the way today
- Stakeholders want answers faster than a ticket queue allows.
- Generic AI tools guess column names and business definitions, and the guesses look plausible.
- Warehouse costs spike when exploratory queries scan whole tables.
What DataLens AI does for marketing analysts
Your schema and definitions
DataLens reads table schemas and sample values, and uses schema notes you write so "revenue" or "active customer" mean what your team means.
Read-only, reviewable SQL
Only SELECT queries run. Every answer shows its SQL in place, and Verify re-runs it live and flags if the numbers changed.
Cost guardrails
Dry-run estimates before every query and a configurable scan cap that blocks expensive queries before they bill.
Your choice of model
Bring your own key for Gemini, Claude, GPT, Grok, Kimi, NVIDIA or OpenRouter, or point DataLens at your own OpenAI-compatible endpoint.
Questions you could ask on day one
- “Write and run the query for weekly new-customer CAC by channel, using our orders table.”
- “Which tables hold refunds, and how do they join to orders?”
- “Estimate the cost of scanning the last 12 months of events before running it.”
- “Turn this question into a scheduled weekly report.”
Which sources does it connect?
Try it on your own data
Sign in with Google to start a 7-day free trial in your own private project — no credit card. Bring your own model key (or request managed models), connect a source, and every answer arrives with its SQL.