Getting Started with Affinda

This guide will walk you through the initial steps to create a new Document Type model in the Affinda platform, allowing you to experience how easy it is to set up a new model for any document type that you need to extract data from.

1. Create a Workspace

When you first access your account, you'll be prompted to create a Workspace. This is the container where all documents are processed and organized. Each Workspace can include multiple Document Types, and you can optionally set up restricted access to limit usage to specific team members within your Organization.

2. Upload documents

Start by uploading a document to begin the setup of your first Document Type and to kick off data extraction.

Upload options include:

  • Directly in the app (recommended for getting started)
  • Emailing an attachment to your Workspace’s unique address
  • Programmatic upload via API

For more information about uploading documents, see Document Ingestion Options.

3. Create a new Document Type

After your first document upload, you’ll be prompted to create a Document Type—a category of documents like invoices, resumes, or payslips, from which you want to extract data.

The platform suggests values for Document Type Label and Description based on the uploaded file. Edit these to reflect the full range of formats this type should cover. These fields are essential for document classification, working alongside Model Memory to improve future accuracy.

See Creating a new Extraction Model for a detailed guide on creating a new Document Type, or see Document Classification for more information about how classification works.

4. Configure Fields

Now, define the fields you want to extract from this Document Type.

You can configure fields in two ways:

  1. Suggested Fields – Affinda recommends fields based on your document. Add, remove, or adjust as needed.
  2. Manual Selection – Draw boxes directly on the document to define new fields. Affinda will auto-suggest names and types.

Field configuration options include:

  • Label, description, and field type
  • Field type-specific settings
  • Text transformations
  • Additional formatting preferences

See Field Formatting for a full explanation of field configuration options.

Once you're satisfied, click "Finish Configuration" to move to the validation interface.

5. Confirm document

In the final step, review and confirm the data extracted from your document. Confirmed documents are added to Model Memory, providing high-quality reference examples for future predictions.

Why this matters: Confirming ensures model accuracy and reduces manual work over time. Incorrect confirmations can negatively affect future results.

See User Validationand Overview of Information Extraction Models for how this improves extraction accuracy.


Next steps

Once you have created your first Document Type, it is time to start thinking about moving your document processing into production.

Validate more documents

Once you have created your Document Type, we recommend uploading and validating more documents to improve performance where needed.

Confirmed documents are added to Model Memory, creating a set of reference data and documents that Affinda's models use to enhance accuracy over time. Combined with our fingerprinting algorithm, this means that the model is presented with relevant example documents when processing newly uploaded documents. As a result, it is important that any data from confirmed documents is accurate, or else future predictions may suffer from poor data.

Create more Document Types

If you need to extract data from multiple documents, follow steps 2-5 in the above to add new Document Types.

Configure your Workflow

Affinda's Platform offers a configurable workflow that can be adjusted to meet your requirements. A typical use case will have documents follow a similar flow, however, the settings within each step of the process can be adjusted (e.g. what classification model is used, how does machine validation work, etc.)

See Workflow and the relevant sections in this Product Guide for each step for more detailed information.

Invite users

Administrators can invite colleagues to their Organization and assign predefined roles and permissions. Once accepted, users can collaborate, manage documents, and contribute to workflows based on their assigned access level.

To invite new Users to your Organization, navigate to the 'Members' settings from your Organization homepage.

For more information about user roles and permissions, see User Management and Permissions.

Integrate Affinda

Integrating Affinda is a straightforward exercise using our REST API. There are 3 main points of integration to consider:

  1. Document Ingestion - While documents can be uploaded directly to the app, generally customers will either upload by sending attachments via email or via API.
  2. Data Mapping - To improve validation and the accuracy of the data being sent downstream, your master data from your CRM, ERP or other platform can be ingested by Affinda via API and subsequently be mapped against the extracted data from the document. Learn more about Data Mapping.
  3. Data Export

See our Integration Guide for more detailed information.