The assistant only runs read-only queries through its database tool. It never runs
INSERT,
UPDATE, DELETE, or DROP on its own — data changes go through the table browser or SQL
editor.Chat
Open the chat panel from the connection toolbar. Ask in plain English:- “Which tables store customer order data?”
- “Explain this query and whether it can be optimized.”
- “Find users who signed up in the last 7 days but haven’t ordered.”
Models
The assistant runs on a frontier model automatically — nothing to pick or configure. If the primary model is overloaded, Tamery falls back to another top model to stay responsive.Claude
Anthropic’s Claude (Opus) is the default.
GPT
OpenAI’s GPT is an automatic fallback.
Gemini
Google’s Gemini is also a fallback.
What it can do
Write SQL from English
Describe the data; get a complete query using your real schema.
Explain queries
Paste SQL and ask for a step-by-step explanation.
Fix errors
Share an error (or use the inline Fix button) for a corrected query.
Optimize / refactor
Add a JOIN, introduce a CTE, rewrite a subquery as a window function.
Query your database
Runs SELECTs to check counts or sample data when answering.
Filters from English
Describe the rows you want; get structured filter conditions.
Enhance prompt
Click Enhance before sending to rewrite a vague message into a clearer one, with schema context added. You review the result before it’s sent.Natural-language filters
1
Open the filter panel
In the table browser, click Filter.
2
Describe the filter
Click Ask AI and type, e.g. “orders in the last 30 days over $500 that haven’t shipped”.
3
Review and apply
The AI builds structured conditions. Adjust, then Apply.
Free accounts get a limited number of AI filter uses per month. Paid plans are unlimited. See
Plans.
AI data seeding
Open a table, click Seed, set a row count. Tamery picks generators from column names and types — names, emails, timestamps, and foreign keys referencing existing rows. Rows are previewed before insertion.Plans
See Plans.