Natural language → SQL
Describe what you need. DataZen uses the current database schema as context and streams SQL you can run or edit.
- Schema of the current connection is included automatically
- Streaming output — stop anytime
- One-click run, or send to the editor
- Supports OpenAI, Anthropic, DeepSeek, and custom endpoints
After an error, no more guessing from docs
When a query fails, click "Diagnose". AI combines the error message and schema to pinpoint the cause and suggest corrected SQL — apply it to the editor in one click.
- Structured display: original error + AI analysis + fix suggestions
- One-click apply corrected SQL — no manual editing
- Handles common syntax, column name, and type mismatch issues well
EXPLAIN insights
See the plan, spot the bottleneck, and get plain-language optimization hints.
- Visual execution plan — scan type, rows, cost at a glance
- AI highlights slow nodes and suggests improvements
- Supported on built-in drivers such as PostgreSQL and MySQL
AI Chat sidebar
Chat with full context of the current connection — schema-aware answers, not generic guesses. Streaming replies; reasoning models supported where the provider allows.
- Auto-injects Schema context — no need to paste DDL manually
@references local files (with size and path limits)- One-click extract code blocks as SQL into the editor
Table filters in plain language
"Amount greater than 1000 and status is shipped" — type natural language in the filter box. AI parses it into structured filter conditions and applies them, with column validation and null operators handled automatically.
- Natural language → structured filter conditions, viewable and editable anytime
- Validates against current table column names to avoid typos
- Works seamlessly with sort, pagination, and virtual scrolling
Beyond querying
- Schema documentation: one-click Markdown docs for your database structure
- Connection troubleshooting: when connections fail, diagnose SSL / SSH / timeout settings
- Query history analysis: auto-summarize categories and purposes of past queries
- Configurable prompts: driver-level + user-level overrides; AI responses follow UI language
