Overview
You can upload files to AI Agents, enabling Bizagi's AI to provide quick insights and solutions based on their content. Whether it is extracting text from images, analyzing data from documents, reviewing financial information, or interpreting technical documentation, the AI significantly accelerates decision-making processes. It also simplifies handling large volumes of information or complex data, including handwritten text.
When creating the Prompt for an AI Agent in the Development environment, configure it to include file usage. In the Production environment, files can either be uploaded manually or generated automatically by Bizagi using existing features, such as Document Templates. The configured Prompt will guide the AI Agent on how to process and utilize these files.
An AI Agent can process one or multiple files, offering flexibility and efficiency in managing the required information.
Steps to configure files as inputs
To configure files as inputs in your AI Agent, follow these steps:
1. Create a new AI Agent
For detailed information, refer to Create AI Agents.
2. Add the input file
Whether you choose to create your AI Agent using a Template or build it from scratch, you can configure it to process files by clicking the Add Input File button during the setup. This allows you to specify the files the AI Agent will handle.
In the Prompt section, define the actions you want the AI Agent to perform on each input file. For better results, clearly specify in the Prompt what the AI Agent will receive, what each file represents, and the desired output.
When setting up a process to use an AI Agent, you must define the input file source through rules mapping. This step applies whether the file is manually uploaded by the end user or generated as a Document Template from a previous process step.
|
You can use rules and Document Templates to generate files containing text that will be used as the output of the AI Agent. |
3. Name the input file and select its type
Provide a name and select a file type for the input file.
The following file types are supported:
•Text (.md, .txt): Basic text files. The content is extracted as plain text.
•Image (.jpeg, .jpg, .png): Image files are processed as byte arrays and sent to the GPT for analysis.
•Portable Document Format (.pdf): PDF files are processed as PNG images, converted to byte arrays, and sent to the GPT.
•Word (.doc, .docx): Word files are processed using Aspose to extract their text content as a String, which is then sent to the GPT. Embedded images within Word files are not processed.
•Excel (.xls, .xlsx): Excel files are exported as CSV files and sent as text to the GPT. Embedded images within Excel files are not processed.
|
•File size comply with the limits defined in Automation Services. •To use files as inputs, the AI Agent must be configured to use the GPT-4o model. •Files that are password-protected or encrypted are not supported. |
4. Add the input file
After clicking Create, the named input file will appear in the Input Files section.
|
•AI Agents do not support Collections. However, you can handle multiple files by creating a single input file and using Merged Actions to combine multiple files, concatenating them with a rule. •You can upload multiple images and files to analyze different file types and extract insights using a single AI Agent. |
5. Save the AI Agent
Once configured, the new AI Agent will appear in the selected Category.
Adding input files to an existing AI Agent
To add input files to an existing AI Agent, navigate to the Expert View, and select the AI Agent that you want to edit. Then, click the Edit Agent button located in the ribbon.
When the editing view opens, create a new input file as described above.
|
•AI Agents can be executed either from an Activity Action or through Form Actions. •You can configure multiple AI Agents within a single process or include multiple AI Agents in a single form. |
Last Updated 11/25/2024 4:57:18 PM