The Generic Indexing action takes the fields produced by the AI Generic Extraction action and writes them into your Record Type fields. It is the general-purpose sibling of Invoice Indexing: instead of a fixed set of invoice fields, every mapping is one you define by name, so it works for any document type.


Use this action after AI Generic Extraction. That action stores each extracted value on the Record as temporary data — header fields as Index_<name> and line-item fields as IndexLine_<name> — and Generic Indexing copies those values into the real Record fields you select.


WORKS WITH: the AI Generic Extraction action. Run that action earlier in the same process so the Index_* / IndexLine_* data exists for this action to map.

How it differs from Invoice Indexing

  • Invoice Indexing offers named rows for the standard invoice fields (Vendor, Invoice Number, Amount, and so on) plus optional custom mappings.
  • Generic Indexing has no standard rows at all — you map every field yourself, or let Auto-Map do it. Each manual mapping pairs an AI Field Name (the name you gave the field in AI Generic Extraction) with the Record field to write it to.


Auto-Map Fields

Because AI Generic Extraction names each field after the Record Type field it came from, this action can apply them for you. Turn on Auto-Map Fields and, for every Record Type field that has an extracted value, the action sets that field automatically — Field = Index_Field for header fields and Field = IndexLine_Field for each line item. Fields with no extracted value are left untouched.


This covers the common case with no configuration: extract with a Record Type selected, then run Generic Indexing with Auto-Map on. You only need the mapping rows below when a field's name differs from the extracted name, or when you want to apply Clean Up. Those rows run after Auto-Map, so an explicit mapping overrides the automatic one for that field.

Auto-Map deliberately skips two kinds of field so it never disturbs values that are set on purpose: the reserved RecordType field, and any field that has a Filter Value (or Filter Value To) configured on the Record Type. Filter values are search and routing constraints, not extraction targets, so they are left exactly as configured. If you do want a filtered field populated from extraction, map it explicitly with a mapping row below — the exclusion applies only to Auto-Map.


Auto-Map of line item fields requires an Enterprise license, the same as the Line Item Fields tab.


Exclude From Auto-Map

Auto-Map is all-or-nothing by design, but you rarely want the exception to be "turn it off and list all twenty fields by hand". Each mapping tab has an Exclude From Auto-Map box, shown whenever Auto-Map Fields is on. Enter the Record Type field names you want Auto-Map to leave alone, one per line. Auto-Map then applies every other field as usual and skips these, leaving whatever value the field already has.


Names must match the Record Type field name exactly, although capitalization does not matter. Wildcards and partial names are not supported — Invoice will not match InvoiceDate. The Header Fields tab excludes header fields and the Line Item Fields tab excludes line item fields, so a field name entered on one tab has no effect on the other.


The exclusion applies to Auto-Map only. If you exclude a field here and also add a mapping row for it below, the mapping row still runs — that is the intended way to take a field out of the automatic pass and then handle it your own way, such as pulling it from a differently named extracted field or applying Clean Up to it.


You do not need to list the reserved RecordType field or any field that has a Filter Value configured on the Record Type. Auto-Map already skips those.


Options

The mapping rows are grouped into two tabs. Use them alongside or instead of Auto-Map.


Header Fields

Each row maps one extracted header field to a Record field:

  • AI Field Name — The name of the field as you defined it in the AI Generic Extraction action's Extra Instructions (for example PatientName). This is matched to the Index_<name> data on the Record.
  • Record Field Name — The Record Type field to write the value into. Leave blank to skip the mapping.
  • Clean Up — Optionally normalize the value as it is written:
    • No Clean Up — write the value as-is.
    • Currency — apply [CLEANCUR()] to strip symbols and separators to a plain number.
    • Date — apply [CLEANDATE()] to normalize the date.


Use Add Header Field to add rows, and the duplicate and delete icons to manage them.


Line Item Fields

The same mapping, applied to each line item the extraction produced. AI Field Names here are matched to the IndexLine_<name> data and written once per line.


ENTERPRISE: The Line Item Fields tab is available on Enterprise-licensed sites. Header field mapping is available on all AI-licensed sites.


TIPS

  1. Start with Auto-Map Fields on and no mapping rows. If your extraction used the Record Type's own field names (the default when a Record Type is selected), that alone indexes the Record. Add rows only for exceptions.
  2. Use Exclude From Auto-Map when Auto-Map is right for almost every field but wrong for one or two — a field a person fills in, a value another workflow action already set, or a field you want mapped with Clean Up instead. Listing the few exceptions is much less work than turning Auto-Map off and mapping every field by hand.
  3. The AI Field Name must match the name you used in AI Generic Extraction exactly — if the extraction instruction said "capture the policy number as PolicyNumber", enter PolicyNumber here.
  4. Put per-document values on the Header Fields tab and per-line values on the Line Item Fields tab. A field extracted per-line will not be found by a header mapping, and vice versa.
  5. Use Clean Up to convert AI text into typed values — Currency for amounts, Date for dates — so downstream searching, sorting and totals behave correctly. To apply Clean Up to a field that Auto-Map already handles, add a row for it — the row overrides the automatic mapping.
  6. If a mapped field comes through empty, confirm the extraction actually produced it (check the workflow log for the Index_* / IndexLine_* values) and that the AI Field Name and header/line placement match.