Getting Started

Workspace Dashboard

This is the main Forge workspace screen that appears after login. From here you can view existing agents, search for agents, and start creating a new workflow.

Key areas on this screen:

• Home — Returns you to the main workspace dashboard.
• Library — Opens the resource or document library.
• All Agents — Shows all agents available in the workspace.
• Models — Opens the model configuration page.
• Audit Logs — Displays activity and system logs.
• Settings — Lets you manage integrations, users, and workspace settings.
• New Agent — Starts the process of creating a new agent.

Begin here whenever you want to create a fresh automation workflow. Click New Agent in the top-right corner to get started.

---

Models Configuration

Before creating an agent, models need to be configured.

This screen is used to create and manage language models inside Forge. Models configured here become available across all agents in the workspace.

The model setup form includes the following fields:

• Name
• Model Name
• Provider
• Host URL
• API Key
• Additional Params (optional)

This page is generally used by an administrator or technical user responsible for connecting external model providers. Typical tasks include:

• Adding a new model provider
• Configuring API access
• Making a model available for agents across the organization
• Testing the model connection before saving

If models are already configured for the workspace, end users typically do not need to make changes here.

---

Create a New Agent

This screen appears after clicking New Agent. Fill in the following:

• Agent Name — A clear, specific name so the agent is easy to identify later.
• Description — A short explanation of what the agent is intended to do and what kind of documents it will process.

After entering the details, click Create Agent.

Once the agent is created, click the + icon next to the File field to define the fields Forge should extract from your documents.

---

Uploading Files

Google Drive Integration

You can connect Google Drive to enable automatic file fetching from a linked folder.

To connect Google Drive:

1. Open Settings.
2. Stay on the Integrations tab.
3. Click Connect to Google Drive.
4. Sign in with the required Google account and grant access permissions.

Once connected, link a folder to an agent. Forge will automatically fetch and process any files added to that folder, removing the need to upload manually.

The Settings screen can also be used to add new members, manage people in the workspace, and control organization-level access.

Manual Upload

To upload files directly from your system, use the Upload Files option and either click to browse or drag and drop files onto the upload area.

---

Field Creation

Fields define what information Forge extracts from your documents. Forge supports six field types:

Single Select — Returns only one response from a set of specified options.

Multi Select — Returns one or more responses from a set of specified options.

Collection — Used for nested structured data.

Text — Used for free-text outputs such as names, titles, or summaries.

Number — Used for numeric outputs. The Custom Data Format option lets you define the precision the model should follow, and you can enable a thousands separator for comma-formatted values.

URL — Used for links such as LinkedIn URLs or website addresses.

Assigning a Model to a Field

You can assign a specific model to each field directly from the field settings.

Choose Source

The Choose Source option lets you add context to a field by selecting any file from the library (shown in the dropdown). This gives the model additional reference material when extracting values for that field.

---

Triggering Evaluations

You can trigger an evaluation in two ways:

Field Level — Select one or more columns to run extraction for those specific fields across all files.

Asset Level — Select one or more rows to run extraction for those specific files across all fields.

---

Agent Output & Results

This screen shows the extracted results for your agent. Each row represents one processed file, and each column represents a field defined in the agent.

Use this screen to:

• Review extracted values for each file
• Compare results across multiple documents
• Verify whether the output is correct
• Export the processed data

Views

At the bottom of the results table you will see multiple views — for example Main and Need Human Review — and the option to create a New View.

Views help you organize processed data and separate records that may require manual checking.

Creating a View — View creation is supported for Single Select and Multi Select field types. You can create a view from the field settings or from the bottom bar, then link it to a specific option to automatically filter assets into that view.

---

Full Workflow Summary

Here is the complete end-to-end journey in Forge:

1. Start on the Home screen.
2. Click New Agent and enter the agent name and description.
3. Define the fields the agent should extract.
4. Configure or select an existing model.
5. Connect Google Drive in Settings (or upload files manually).
6. Let Forge fetch and process files automatically.
7. Review the extracted results in the agent table view.