Tailored models are versions of one of Affinda's off-the-shelf models (e.g. invoices, receipts, resumes) that have been trained on documents specific to a customer's use case to get even better performance. The most common use case of a tailored model is within the Accounts Payable space. Often, a customer's supply chain will have a small number of suppliers that encompass a large proportion of their invoice volume. With a tailored model, an initial focus on these suppliers can greatly reduce the amount of processing time required. The 'long tail' of suppliers that are of lower volume will continue to improve in accuracy as they continue to get validated and fed back into the model for training.

How tailored models work

Tailored models work by using the base 'off the shelf' model as a foundation. The base model will be trained across a large and varied set of documents and thus will perform well 'out of the box' on most document formats of the same type. A tailored model then adds a layer that is specific to the document formats seen by that user account. After seeing a small number of documents of each document format type, the model can quickly learn and see higher accuracy, reducing (or eliminating) the time users need to review a document.

Using tailored models

While using tailored models as part of your document automation solution can be a powerful tool to deliver even greater efficiencies, they are not appropriate for all use cases and the number and configuration of the tailored models needs to be considered to ensure no regression occurs over time.

Decision 1: Do I need a tailored model?

The decision on whether a tailored model is appropriate comes down to:

Document type

Tailored models will work well when there is expected to be some level of consistency in the document formats received over time so that the tailored model can start to ‘learn’ formats that it sees consistently.

• For example, this is why invoices are a particularly good use case because of repeat supplier relationships, whereas receipts may be less appropriate as these transactions are typically more once-off in nature.

Workflow & feedback loop

An element of human review needs to be included in the final solution using Affinda’s validation UI to ensure that we are able to receive back the corrected data in the right format for training

• For example, users do not currently have the option to correct resume data, and as such a tailored model is not appropriate.

If it has been determined that a tailored model is appropriate for the use case, a decision is needed about how to implement this.

Decision 2: How many tailored models are required?

For most customers, a single tailored model per document type is likely to suffice. However, multiple tailored models may be required for your Organization for the following reasons:

Validation team(s)

• Typically, we recommend a tailored model per validation team
• This is due to different teams potentially validating documents differently or inaccurately
• Enterprise customers typically only have a single AP team processing documents and as such a single tailored model often works
• However, for software platforms delivering the solution to multiple end-users, it is important to segregate the tailored models to ensure one customer doesn’t see a regression due to another customer’s validations

Field requirement(s)

• For consistent learning, tailored models need consistent data to be supplied
• If different fields are required to be returned (e.g. some fields not relevant, additional custom fields), separate tailored models should be set up for each ‘Collection’ of field settings

Document type(s)

• Tailored models will often perform better if they are required to generalize across a smaller set of document formats, even within one overarching document type (e.g. invoice)
• Performance may be optimized by creating different tailored models by industry, jurisdiction, or for specific customers/use cases
• Understanding the above dimensions helps determine how many tailored models are required and what this means for how to configure your Organization within the Affinda app.

Configuring your Organization for tailored models

We recommend that users set up a single Collection that can be mapped 1:1 with a tailored model. This simplifies the setup and configuration process so that the risk of model regression due to data inconsistencies is lower.

Setting up tailored models

To discuss further whether tailored models are appropriate for your use case and to understand further how to best configure your account to take advantage of them, contact us to set up a discussion so that we can understand your use case better and provide some recommendations.