Shu Qingzhou Agent
Shu Qingzhou Agent is a localized Excel data processing intelligent agent launched by Beijing Qiance Shuzhi Technology Co., Ltd. It supports access to local large models, data cleaning and multi-table association, and automatically saves the generated code as scripts for offline reuse.
Updated: 2026-10-26
What is Shu Qingzhou Agent?
Shu Qingzhou Agent is a localized Excel data processing AI agent launched by Beijing Qiance Shuzhi Technology Co., Ltd. Its official website positions it as "a localized AI agent that can process Excel data with one sentence". It has two core selling points: running the large model locally or through the Silai Flow interface to keep data on the local machine; automatically saving the generated code as a script to avoid repeated consumption of Tokens during repeated processing.
Who is it for?
- People in finance, operations, procurement and other roles who work with Excel every day: codify repetitive tasks such as data cleaning, multi-table merging and summarization into scripts.
- Teams with data compliance requirements who cannot upload spreadsheets to the cloud: Local model deployment is its main differentiator compared to pure cloud-based tools.
- People who need to regularly generate analysis reports: Use custom report templates to reduce manual formatting.
- Analysts working with million-row tables: Officially supports files of 500MB in size and chunked parallel computing.
Key features
6Local Large Model Access
The official states that it supports local deployment of large models, with data processed locally without being uploaded to the cloud, enabling offline operation; the page lists that models such as DeepSeek, Qwen, Kimi, GLM, Doubao, MIMO, and Tongyi Qianwen can be connected.
Intelligent Data Processing
Supports Excel, CSV and other formats, enabling data cleaning, transformation, analysis and visualization. According to the official description, it can automatically identify null values, outliers and duplicate data, and write Python programs in the background for processing. It also supports batch processing and multi-table association.
Script Management
The processing code generated by the large model is automatically saved as a script. The official states that it can be reused even when offline, and repetitive data processing logic no longer consumes Tokens or points, allowing unlimited local reuse.
Data Analysis Report Generation
Automatically generates professional data analysis reports, supports export in various formats. The official states that it provides intelligent content generation and allows customization of report structures and styles.
Large-Scale Data Support
The official states that it can efficiently process files with a data volume of up to 500MB (millions of rows). It automatically writes Python programs in the background for block-based parallel computing, enabling one-click data merging, splitting, association and filtering.
Semantic Custom Query
Supports natural language semantic custom queries. The official claims that it can accurately match data according to industry terms and custom rules, truly enabling data query in plain language.
How to use Shu Qingzhou Agent
- The official process has three steps: first configure the model (either a local large model or a Silicone Flow interface), then select a function (data processing, script management, or report generation), and finally converse to process and download the results.
- It is recommended to run through the entire process with an anonymized small table for the first time.
- The performance of the local model depends on the capabilities and hardware of your machine.
- The official does not guarantee equivalence to cloud-based large models, and important conclusions should be manually verified.
- Script reuse is the key to saving money: first confirm that the logic of the generated script is correct before reusing it repeatedly.
- otherwise, incorrect cleaning rules will be amplified in batches.
- It is officially marked as supporting Windows 10 and above systems.
- Google Chrome or Microsoft Edge is recommended for access, and users should confirm compatibility with other systems first.
- The official website offers a lifetime buyout option, with a limited-time discount price of 89 yuan and an original price of 149 yuan, allowing permanent use with a one-time payment.
- the price and authorization scope are subject to the real-time content on the official website, and the authorization terms should be confirmed before commercial use.
Traffic & popularity
Data month Sep 2026
Shu Qingzhou Agent receives about 6.2K visits per month.
Monthly visits
7 monthsEstimated from third-party traffic data.
Similar tools
ExcelMaster.aiExcelMaster.ai is a spreadsheet-oriented AI agent that can generate complex formulas using natural language, write and explain VBA macros, perform cell-level operations, and includes free tools such as built-in PDF/Word/Excel conversion functions.
UnslothOpen-source local AI workstation: download models and chat directly on the desktop, fine-tune models without writing code, locally generate and edit images and videos, and connect coding agents to your own GPU.
AlgForce AIAlgForce AI is an AI data analysis and report generation platform. You can upload tables to generate data dashboards and analysis reports. The official claims that it can one-click generate multi-page reports, and it also provides MCP for Agent to call.
OllamaOpen-source large model local runtime: one-click to launch the model, and prompts never leave the local machine, yet can also connect to the cloud. Suitable for developers who run models locally and are concerned about data leakage, it supports installation on all three desktop platforms.
RowSpeakRowSpeak is an AI data analysis platform for sources such as Excel, CSV, and PDF. It can analyze tables, clean data, generate charts and interactive dashboards with natural language, and support converting images/PDFs to Excel.
CertantCertant is a white-label AI platform built for consultancies and systems integrators. It offers two models: fixed licensing for customer-provided hardware and cloud-based credits. The company claims it has already completed production deployments covering 120,000 pages.