IXP（预览版）

备注：

使用编码智能体构建 IXP 项目现已进入公开预览阶段。在正式发布之前，功能和命令可能会发生变化。

IXP, also known as Intelligent Xtraction and Processing, turns unstructured documents — invoices, receipts, forms, contracts — into structured data your automations can use. With a coding agent, you point IXP at sample documents, review the data it predicts, and improve the project until it extracts reliably.

您构建的内容

You build an IXP project. The agent creates the project from your sample documents, IXP suggests a taxonomy and predicts field values, and the agent reviews those predictions against each document — confirming what is correct and refining the extraction instructions where it is not. The result is a published project your RPA and API workflows can call to read documents.

This suits document-heavy work: reading totals and line items from invoices, or pulling fields from forms and contracts.

背后的技能

This work is powered by the uipath-ixp skill, which manages IXP projects — creating projects, reviewing predictions, improving extraction prompts, and publishing projects. You can see it alongside the rest in the Skills catalog.

典型的启动提示词

第一个请求可能如下所示：

“从此文件夹中的发票创建一个 IXP 项目，提取发票编号、日期、总计和明细项目，然后查看预测并告诉我提取分数。”

From there you refine the project conversationally — improving the instructions for low-scoring fields, then publishing it when the scores are strong.

更深入了解

For the full picture of IXP — building, deploying, and consuming projects — see the IXP documentation.