About AskTable
Overview
Guide · Module: product · Users · Deployment: Not specified · Confirmed product versions: Not specified · Source: Existing documentation
AskTable is a data question-answering platform: connect your databases and spreadsheets so people can get answers without writing SQL, with every number traceable to its source.
AskTable's core workflow: natural language → real SQL → real data → traceable results.
Why answers are trustworthy
| Basis for trust | Explanation |
|---|---|
| Actual execution | Data comes from SQL queries or Python calculations. AI explains the results rather than inventing numbers. |
| Inspectable process | View the SQL or Python code that ran to understand how data was queried and calculated. |
| Honest answers | When no data source is available, relevant data cannot be found, or its meaning is unclear, the agent says so. It does not fabricate results or present uncertain data as conclusions. |
| Active clarification | With user clarification enabled, the agent pauses to ask about unclear time ranges, metric definitions, or other conditions, then continues after confirmation instead of guessing. |
Read the full explanation of how it works.
Three capabilities
| Entry point | What it does | Who it is for |
|---|---|---|
| New conversation → Conversations | Answer a specific question with tables and charts, then ask follow-up questions | Business users in operations, finance, HR, sales, and similar teams |
| Analysis canvas → Analysis canvases | Break analysis into steps for attribution, drill-downs, and multidimensional comparisons | Data analysts and roles that need in-depth analysis |
| Data dashboards → Data dashboards | Build an automatically refreshing page through conversation to track metrics over time | Anyone who needs ongoing metric monitoring |
Where your data stays
- Self-hosted by default. Start the full stack with Docker Compose. It can run entirely offline on your private network, keeping raw data, metadata, and logs within your network.
- Models can also be self-hosted. Connect an internal OpenAI-compatible or Anthropic-compatible service without sending questions to the public internet.
- Traceable access records. Each question's execution trace, sources, and user ratings are retained in the agent's audit logs.
Supported data sources
The interface groups more than 30 engines into Files, Databases, and Big data. See Connect data sources for the full list.
Chinese domestic and Xinchuang databases have native support rather than relying on generic drivers:
| Category | Supported sources |
|---|---|
| Chinese domestic centralized databases | Dameng DM8, KingbaseES V8, GBase 8c V5, Xugu V11/V12, YashanDB 23.2, MogDB 5.0.0 |
| Chinese domestic distributed / cloud-native databases | GaussDB (primary/standby edition), GaussDB(DWS) 8.x, GBase 8a V9, TDSQL (MySQL / PostgreSQL), OceanBase 4.2.2, TiDB 8.1.0, PolarDB (MySQL / PostgreSQL), AnalyticDB (MySQL / PostgreSQL) |
| Big data and warehouses | StarRocks, ClickHouse, Doris, SelectDB, Hive, MaxCompute, Hologres, Greenplum, Databend, Google BigQuery |
| General-purpose databases | MySQL, PostgreSQL, Oracle, SQL Server |
| Files | Excel / CSV, built-in Workbook |
Who it is for
- Business users: ask data questions yourself without waiting in the data team's queue
- Data analysts: let business users handle recurring questions themselves so you can focus on deeper analysis
- Data administrators: connect data sources, maintain field semantics, and configure access scopes
- System administrators: manage self-hosting, model configuration, licensing, and integrations
Deployment options
| Option | Description |
|---|---|
| Self-hosted (default) | Start the full stack with Docker Compose. It can run entirely offline on your private network, keeping data within your security boundary. See Self-hosting |
| SaaS | Extends the self-hosted foundation through plugins, including billing and multitenancy. Use “Go to console” at the top right of the site |
Fit AskTable into your team's workflow
| Integration | Use case |
|---|---|
| Feishu bot | Ask data questions directly in private or group chats |
| Web embedding | Embed data Q&A in your own system, including access for visitors |
| Scheduled analysis reports | Automatically analyze data and deliver conclusions on a schedule |
| Command-line tool | Scripts and automated workflows |
| MCP | Connect AI clients with 43 tools |
| OpenAPI | Call directly from business systems |
Ready-to-use resources
- 9 data agent templates: Store Operations Analyst, E-commerce Data Monitor, Financial Data Analyst, Market Insights Analyst, Supply Chain Monitor, User Growth Analyst, Executive Data Assistant, Business Traffic-light Analyst, and Data Quality Guardian. See Agent overview
- 11 analysis skill templates: anomaly detection, trend forecasting, stress testing, metric drill-down, report orchestration, comparison analysis, attribution analysis, metric interpretation, data quality checks, business-language generation, and periodic analysis. See Analysis skills
Next steps
- First time here: Ask your first question
- Connect your own data first: Connect data sources
- Learn how it works internally: How AskTable works