Using Claude API to extract the entities from a email document. For my use case we need to extract the starting and ending positions of the extracted values as well. Tried multiple prompts to get the indexes. Have a plan to do string matching, but there is some prompt for certain values. For example,
”’
Text: “Subject: sample subject” For this subject value we can easily do the string matching.

Text: “Payroll: $123 (payroll)” For this kind of values where there call Payroll and category call payroll, for this kind of scenario, string matching wont work.
”’
Anyone faced similar kind of an issue ? Any idea how can we get the indexes for the extracted value.