数以轻舟Agent

Editor Review

Start by saying in one sentence whether the sheet needs cleaning or a summary, instead of writing formulas by hand. Local scripts store the generated program so the same logic can run again offline, without spending quota on that repeated step. Report export produces documents from finance, market, or operations templates. The model can use a SiliconFlow connection or a local model, and the page stresses that data is processed on the machine. It fits analysts who clean the same kind of workbook repeatedly and do not want the detail rows uploaded. Macs are not on the operating-system list the page states.

AI Tools Navigator Editorial TeamUpdated: 2026-10-09

What is 数以轻舟Agent

Shuyi Qingzhou, from Beijing Qiance Shuzhi Technology, is a local spreadsheet agent. Users describe cleaning, transforms, summaries, or cross-sheet joins in plain language. The tool writes a processing program on the machine and stores that code as a script that can run again. Data can stay with a local model. The page lists connectable models from Tongyi Qianwen, DeepSeek, Moonshot, Zhipu, Doubao, and Xiaomi. It can also build analysis reports for finance, market, or operations needs. The client targets Windows and UnionTech OS, and the purchase is a one-time license.

数以轻舟Agent Key Features

Sheets from plain language

Describe in plain language how to clean blanks, outliers, and duplicates, or how to transform, group, and join sheets. A processing program is generated in the background.

Reuse scripts offline

Code written by the model is saved as a script. The same logic can run again while offline, without another model request for the repeated step.

Process with a local model

It can connect models from Tongyi Qianwen, DeepSeek, Moonshot, Zhipu, Doubao, and Xiaomi, and process data on the machine.

Export an analysis report

Builds reports for finance, market, or operations needs, then exports them after the structure and style are adjusted.

Chunked processing for large files

For larger sheet files, it writes a program in the background and computes in chunks for merge, split, join, and filter jobs.

数以轻舟Agent Key Advantages

Cleaning, summaries, and cross-sheet joins are described in plain language, without memorizing spreadsheet functions first

Generated processing code is stored as a script and can repeat the same logic while offline

A local model can be connected, and the page says detail rows are processed on the machine rather than uploaded by default

Reports can be restructured for finance, market, or operations scenarios before export

数以轻舟Agent Use Cases

Clean the monthly finance sheet again

A finance user describes blank-value and format rules in one sentence, then keeps the first script to run on the same kind of sheet next month.

Summarize several operations sheets

Operations hands several detail sheets to a join and group step and gets the summary without writing cross-sheet formulas.

Write a report on an internal network

An analyst configures a local model, generates a market or operations report without uploading the detail rows, and exports it.

Frequently Asked Questions

▶Are the sheet data uploaded?

The site presents local processing as the main path: a local model can be connected, detail rows are computed on the machine, and offline running is described. A model connection can also go through SiliconFlow. Where the key is stored, and when the network is still contacted, is pointed to the FAQ rather than spelled out on the homepage. For repeated jobs, a saved script can run again while offline.

▶Does the same cleaning call the model every time?

No. After the model writes the program the first time, the code is saved as a script. The site says the same logic can be reused offline, without spending quota on that repeated step. That suits monthly finance or operations sheets with a similar structure. A new rule still needs a new description and a new script.

数以轻舟Agent

Editor Review

What is 数以轻舟Agent

Basic Info

Best For