Observe words and count actions.

What is AskTable, which involves observing words and numbers?

AskTable is an AI intelligence tool that enables business professionals to query and analyze corporate data using natural language.

The current product is operated by Hangzhou Memory Future Artificial Intelligence Co., Ltd., and it focuses on addressing issues related to data parameters, permissions, and the analysis of accurate data.

Main functions of AskTable

• Query databases and tables using natural language.
• Automatically generate queries, summaries, and charts.
• Continue to ask questions and delve deeper based on the results.
• Metrics for precipitation, field notes, and business documentation.
• Implement data access permissions based on identity.
• Configure row-level, field-level, and masking rules.
• Used in Web, Lark, API, MCP, and CLI.
• Package high-frequency analysis as skills and digital employees.
• Supports SaaS, on-premises, and open-source deployment.

What is the difference between it and traditional BI?

Traditional BI typically involves designing fixed reports in advance, whereas AskTable places more emphasis on business users asking questions directly and pursuing further inquiries.

It doesn’t merely translate problems into SQL; it also takes into account semantic configurations, metrics definitions, and user permissions.

• For temporary issues, there is no need to wait for creating a new report.
• The answer can be returned in the form of tables, charts, and explanations.
• The same question can yield different results depending on the user’s permissions.
• At the level of business knowledge, inconsistencies in indicator definitions are reduced.
• Traditional BI is still suitable for fixed dashboards and complex visual designs.

Which data sources are supported?

AskTable can connect to databases, data warehouses, files, and certain business platforms.

• MySQL and PostgreSQL.
• ClickHouse and StarRocks.
• Databases such as OceanBase and TiDB.
• Excel and CSV files.
• Lark Multi-dimensional Table.
• Business systems such as ERP, CRM, and Yonyou.
• Custom data sources can be extended through DAP.

The specific version support range and driver requirements shall be based on the deployment documentation and the console.

Natural language question answering

Users can directly ask questions regarding sales, inventory, finance, membership, or operations.

• Query the key metrics for a specific time period.
• Break down the data by region, channel, and product.
• Drill down into and attribute abnormal fluctuations.
• Compare the performance of different cycles or teams.
• Generate tables, charts, and textual conclusions.
• Ask about the missing conditions and expand the filtering criteria.

Semantic layer and business definitions

The semantic layer is used to store the meaning of fields, formulae for metrics, synonyms, business documents, and analysis preferences.

These configurations enable AI to map business language to the correct data tables and calculation rules.

• Add a business name to the technical field.
• Define the criteria for metrics such as revenue, activity, and conversions.
• Record the time range and filtering criteria.
• Add common abbreviations and synonyms for departments.
• Associated binding systems, instructions, and data dictionaries.
• Set analysis preferences for different scenarios.

Permissions and data governance

AskTable can limit the range of data before a query is executed, based on the user’s identity, role, and access policies.

• Control the accessible data sources.
• Restrict specific tables and fields.
• Restrict the range of records using row filters.
• Mask sensitive fields such as phone numbers.
• Assign permissions to administrators, members, and visitors.
• Maintain the ability to audit query and management operations.

Incorrect permission settings can still lead to unauthorized access; it is necessary to use a test account for verification before the system goes live.

Charts, Explanations, and Follow-up Questions

The system can organize the query results into tables, charts, and textual descriptions, and it supports further follow-up questions.

• First, check the total amount and trends.
• Break down the differences further by dimension.
• Identify the abnormal time points or business objects.
• Compare the target value with the actual value.
• Summarize the conclusions into a report or action recommendations.

AI-based interpretation serves as an auxiliary analysis tool; the final decision still requires verification of the source data and the definitions of the metrics.

Lark, MCP, API, and CLI

The same set of questioning capabilities can be made available to business teams and developers through various channels.

• The Web console is suitable for configuration and interactive analysis.
• The Lark entry point is suitable for asking for data directly in group chats.
• REST APIs are used to integrate into business systems.
• MCP enables external AI Agents to access data capabilities.
• CLI is suitable for terminals and automated workflows.
• The Python SDK supports both synchronous and asynchronous calls.

AskTable usage tutorial

1. Identify the first high-value business analysis scenario.
2. Choose between SaaS, on-premises, or open-source deployment.
3. Create an organization and set up an administrator account.
4. Connect to databases, data warehouses, or spreadsheet files.
5. Use a read-only account and restrict network access.
6. Add field notes, metric definitions, and business documentation.
7. Create roles and configure table, field, and row-level permissions.
8. Test the accuracy of queries using typical business problems.
9. Verify the SQL, results, charts, and explanations.
10. Create a set of evaluation questions with known answers.
11. Connect to Lark, APIs, MCP, or other entry points.
12. Ongoing audits of errors, changes in permissions, and business parameters.

SaaS plans and prices

As of August 30, 2026, the official pricing page uses the total number of employees in an organization as the basis for calculation.

Administrators, members, and visitors are all included in the count; prices and promotion rules are subject to those on the final settlement page.

Free trial

New users can first utilize the free trial quota to test data integration, natural language queries, and basic analysis capabilities.

• First, connect with masked samples or low-risk data.
• Prepare a set of questions whose answers are already known.
• Test field mapping and metric formulas.
• Check whether error issues will be explicitly rejected.
• Upgrade after confirming the team collaboration requirements.

How to choose a SaaS plan

• The trio package can start with the smallest organizational option.
• Visitors are also included in the count; plan the seating in advance before making invitations.
• For cross-departmental promotion, it is necessary to track the total number of actual users.
• Unlimited questions and answers are part of the rules during the promotion period.
• Highly sensitive data should have a private deployment evaluated.
• For groups of more than twenty people, a corporate plan should be consulted.

Private deployment

Official support is available for private deployment; the data can remain within the enterprise’s internal network, a proprietary cloud, or a designated environment.

The price for privatization requires a commercial quote, and it depends on the environment, the volume of data, the scope of delivery, and the service period.

• It supports connection to internal databases and data warehouses.
• Enterprise roles and row/column-level permissions can be configured.
• It can be combined with model strategies, data masking, and auditing.
• It can be integrated with internal systems and workflows.
• Before implementation, it is necessary to assess the computing power, network infrastructure, and operational capabilities.
• Commercial support, upgrades, and SLAs must be specified in the contract.

Comparison of SaaS, open source, and on-premises solutions

Open-source status

The official team has made available the AskTable all-in-one deployment repository, which is licensed under the Apache-2.0 license.

This repository provides Docker-related files, which enable the operation of UI, REST API, and MCP services.

• It supports deployment on standalone systems, Alibaba Cloud, Sealos, and other platforms.
• During construction, you can choose a domestic Python mirror source.
• Deployment requires configuring the database and large model manually.
• The open-source deployment version does not include commercial implementation and SLAs automatically.
• The full hosting capabilities in the cloud cannot be simply equated with the content stored in a warehouse.

Python SDK

The official Python library for accessing the AskTable REST API is also licensed under the Apache-2.0 license.

• Python 3.8 and later versions are supported.
• Synchronous and asynchronous clients are provided.
• Requests and responses contain type definitions.
• It supports pagination, file uploading, retry options, and timeout settings.
• API keys should be managed through environment variables.
• The call quota and service prices are determined by the deployment scheme used.

API and MCP security

• Create separate API keys for different projects.
• Assign chat or management roles based on the minimum required permissions.
• Do not include high-privilege keys in the frontend code.
• Set up call logging, rate limiting, and exception alerts.
• Check the range of data exposed by the MCP connection tool before making the connection.
• The key is revoked and replaced immediately upon leakage.

How is data accuracy ensured?

Reliable data querying requires the combined effort of data quality, semantic configuration, permission rules, and model capabilities.

• Create a test set that includes the standard answers.
• Check whether the generated query uses the correct tables and fields.
• Verify time, deduplication, and aggregation criteria.
• Show the calculation process for high-risk indicators.
• Establish a clarification process for issues lacking the necessary conditions.
• Update the semantic layer promptly after any changes in the business parameters.

Privacy and user data

According to the privacy policy, users own the AskTable user data that they upload, create, or that results from the operation of bots.

In principle, the platform processes data according to the user’s instructions, but free hosting does not imply an obligation to store the data permanently.

• Users must ensure that the sources of data and the methods used for processing it are legal.
• Accounts, devices, and browsing history may be used for secure operations.
• Personal information generated from operations within the country should, in principle, be stored within that country.
• Deletion, cancellation, and authorization management should be carried out in accordance with the relevant policies.
• Important data must have separate backups.
• When releasing a Bot to the public, users also need to provide their own privacy rules.

Which companies are suitable?

• Teams in sales, operations, and finance that need to retrieve data on a frequent and ad-hoc basis.
• Companies that possess multiple databases and business systems.
• Data departments that wish to standardize metric definitions and access rights.
• Organizations that need to query data directly within Feishu.
• Development teams that aim to provide AI Agents with the capability to work with real data.
• State-owned enterprises and large organizations that require internal deployment and auditing.

Product advantages

• Natural language questioning is integrated with the business semantics layer.
• Permission rules take effect before a query is executed.
• Supports databases, data warehouses, files, and multiple entry points.
• High-frequency analysis can be transformed into skills and digital employees.
• It also offers SaaS, private, and open-source deployment options.
• Officially, REST APIs, MCP, and Python SDKs are provided.

Usage restrictions and precautions

• AI analysis may still produce incorrect queries or interpretations.
• Poor data quality directly affects the reliability of the answers.
• Administrators, members, and visitors are all included in the SaaS user count.
• Unlimited Q&A is part of the promotional period and the rules for reasonable use.
• The privatization price requires a business assessment.
• With open-source deployment, one is responsible for maintenance and security tasks themselves.
• For high-risk decisions, it is necessary to review the source data.

Frequently Asked Questions

What can AskTable, which is used for observing words and numbers, do?

It allows users to query corporate databases and tables using natural language, and returns data, charts, explanations, as well as follow-up results.

Can I use AskTable without knowing SQL?

Yes, sales staff can ask questions directly, but the data team is still required to set up data sources, semantic definitions, and permissions.

How is AskTable SaaS priced?

On August 30, 2026, the monthly fees for groups of three, ten, and twenty people will be 200 yuan, 600 yuan, and 999 yuan respectively.

Does AskTable support private deployment?

Yes, the final price depends on the deployment environment, the volume of data, the scope of delivery, and the service period.

Is AskTable an open-source tool?

The official all-in-one deployment repository is licensed under Apache-2.0, but commercial cloud services and implementation support must be purchased separately.

Does AskTable provide APIs and MCP?

It offers REST APIs, MCP, CLI, and an official Python SDK, which can be used for system integration and calling agents.

Are the analysis results of AskTable always accurate?

Not necessarily; continuous verification is still required through test sets, query checks, metric definitions, and validation of the source data.