Stop Making the Same Correction. Teach AI What to Do Next.

Your team processes an invoice. Most of the information is captured correctly, but the purchase order number needs attention. Someone finds it under “Customer reference,” updates the field, and moves on. Then another invoice arrives from the same supplier. The same correction happens again. Each correction may take only a moment. Repeated across thousands of documents, those moments become a recurring task, and valuable employee knowledge stays locked inside manual work.

Base64’s Teach AI helps turn that knowledge into reusable rules for future documents.

Put your team’s knowledge to work

The people reviewing your documents understand the details that matter to your business. They know which supplier uses an unusual label, when a field needs a specific value, and which documents require a closer look. Teach AI gives teams a way to capture those requirements through rules that define when to act and what to change. A rule can use AI to answer a targeted question and update a field, assign a fixed value, or change a document’s status to Auto-approved, Needs review, or Rejected. This lets you connect document extraction to the needs of the people who use the results.

One recurring issue. One reusable rule.

Consider a supplier whose invoices list the purchase order number as “Customer reference.” With Teach AI, your team can create a rule that identifies invoices from that supplier and asks AI to find the customer reference number and save it in the purchase order field. Once saved, the rule applies to newly processed documents that meet its conditions. The same approach can support other business requirements. For example, a team could configure a rule to mark invoices above a defined amount as Needs review, helping draw attention to documents that require additional oversight. Your team defines the conditions, the action, and where the rule applies.

Create rules where the work happens

A recurring issue is often easiest to explain while looking at the document itself. Teach AI lets users create or update a rule directly from a field or table on the document review page. Teams can also manage rules in their Flow or Custom Model settings. Rules saved to a Flow apply within that workflow. Rules saved to a Custom Model apply wherever that model is used. That flexibility lets teams address a requirement within a specific process or reuse it across workflows that share the same model.

Preview the result before saving

Before saving a rule, users can preview its effect on the current document.

The preview checks whether the conditions match. When they do, it shows the current and updated values so the reviewer can evaluate the result. If a condition fails, the preview identifies the mismatch. Teams can adjust the rule and preview it again before saving it for newly processed documents.

Start with the correction you keep making

A practical place to begin is a familiar question: “What does our team keep fixing?”

Choose one recurring issue, define when it occurs, configure the action, and preview the result. Then evaluate the rule on representative documents and refine it as needed. Over time, this creates an opportunity to turn everyday review experience into more consistent document processing, and give employees more time for work that needs their judgment. Your team already knows what to do with those recurring exceptions. Teach AI helps put that knowledge into practice. Ready to turn recurring corrections into reusable rules? Explore Teach AI and schedule a meeting with our team.