Category: Artificial Intelligence

‘Document assembly’ is the process by which an operator creates an entire document from a variety of component parts and then personalizes that document to meet the needs of the intended recipient.
Included within the scope of the term ‘document assembly’ are how source clauses are:

· created
· neutered, and
· assembled.

One of the first steps in automatically assembling legal documents is identifying terms in square brackets. This is the way that some automated document assembly systems work like Pathagoras. To create a legal document for automation and assembly begins with surrounding key terms such as party names and so on with square brackets eg [William Higgs] or [Higgs Limited]. You get the idea. The idea is that the square bracketed terms will often be repeated throughout a legal document, so it is efficient to define them.

The [Initial Unitholders] have paid the [Initial Sum] to the [Trustee] to establish a trust on the terms of the[Trust Deed].

The terms  Initial Unitholders, Initial Sum, Trustee and Trust Deed are all obviously variables and they would be repeated throughout a contract. A 'variable' is a place holder for personal data. You should strategically place variables within your source clauses where you want that data to appear. Consequently, those variables will also appear in the first draft of any newly assembled document.

At some stage in the assembly we will need to prompt the user to enter those terms.
So lets start by producing a Word Macro that searches for terms in [ ] and produce them in a new Word Document. Once we have the Terms we can then prompt our assembly program to complete the legal contract with those specific terms. This is called an Interview. An Interview is a series of questions asked at the beginning of the final assembly process regarding what you want included/discarded in the final legal document. It consists of questions that are presented in a menu format.
The interview is asked in a single window, with multiple questions being asked. More about the Interview in later blogs.

Here is some code you can enter into a macro in Microsoft Word. Obviously make sure your Word Document has some [ ] terms in it.

Option Explicit

' ===========
' ENTRY POINT
' ===========
Sub ExtractBracketedText_ToTable()
    Dim doc As Document: Set doc = ActiveDocument
    Dim bodyRng As Range: Set bodyRng = doc.StoryRanges(wdMainTextStory) ' main text only

    ' Regex to capture [ ... ] (non-nested), allowing across line breaks
    Dim rx As Object: Set rx = CreateObject("VBScript.RegExp")
    rx.Global = True
    rx.IgnoreCase = False
    rx.MultiLine = True
    ' Group 1 captures the inner text (without the brackets)
    rx.Pattern = "\[([\s\S]*?)\]"

    If Not rx.Test(bodyRng.Text) Then
        MsgBox "No bracketed text found in the main story.", vbInformation
        Exit Sub
    End If

    Dim matches As Object, m As Object
    Set matches = rx.Execute(bodyRng.Text)

    ' Prepare a simple record type via arrays
    Dim n As Long: n = matches.Count
    Dim arrText() As String, arrPage() As Long, arrCtx() As String, arrStart() As Long, arrLen() As Long
    ReDim arrText(1 To n)
    ReDim arrPage(1 To n)
    ReDim arrCtx(1 To n)
    ReDim arrStart(1 To n)
    ReDim arrLen(1 To n)

    Dim i As Long, startInDoc As Long, lengthInDoc As Long
    For i = 1 To n
        Set m = matches(i - 1)
        ' Map back to document coordinates
        startInDoc = bodyRng.Start + m.FirstIndex
        lengthInDoc = m.Length

        arrStart(i) = startInDoc
        arrLen(i) = lengthInDoc
        arrText(i) = CleanOneLine(m.SubMatches(0)) ' inner text only

        ' Page number (guard errors just in case)
        Dim hitRng As Range
        Set hitRng = doc.Range(startInDoc, startInDoc + lengthInDoc)
        On Error Resume Next
        arrPage(i) = hitRng.Information(wdActiveEndAdjustedPageNumber)
        If Err.Number <> 0 Then
            Err.Clear: arrPage(i) = 0
        End If
        On Error GoTo 0

        ' Context snippet around the entire [ ... ] region
        arrCtx(i) = GetContextSnippet(doc, startInDoc, lengthInDoc, 45)
    Next i

    RenderResultsTable arrText, arrPage, arrCtx
    MsgBox "Bracketed-text report created.", vbInformation
End Sub

' =================
' RENDER TO NEW DOC
' =================
Private Sub RenderResultsTable(ByRef arrText() As String, ByRef arrPage() As Long, ByRef arrCtx() As String)
    Dim outDoc As Document: Set outDoc = Documents.Add
    outDoc.Activate

    Selection.Style = wdStyleHeading1
    Selection.TypeText "Bracketed Text Report"
    Selection.TypeParagraph
    Selection.Style = wdStyleNormal
    Selection.TypeText "Generated: " & Format(Now, "yyyy-mm-dd hh:nn")
    Selection.TypeParagraph: Selection.TypeParagraph

    Dim n As Long: n = UBound(arrText) - LBound(arrText) + 1
    If n <= 0 Then
        Selection.TypeText "No results."
        Exit Sub
    End If

    Dim tbl As Table
    Set tbl = outDoc.Tables.Add(Selection.Range, n + 1, 4)
    With tbl
        .Style = "Table Grid"
        .Cell(1, 1).Range.Text = "#"
        .Cell(1, 2).Range.Text = "Extracted Text"
        .Cell(1, 3).Range.Text = "Page"
        .Cell(1, 4).Range.Text = "Context"
        .Rows(1).Range.Bold = True
    End With

    Dim i As Long, row As Long
    row = 2
    For i = LBound(arrText) To UBound(arrText)
        tbl.Cell(row, 1).Range.Text = CStr(i - LBound(arrText) + 1)
        tbl.Cell(row, 2).Range.Text = arrText(i)
        If arrPage(i) > 0 Then
            tbl.Cell(row, 3).Range.Text = CStr(arrPage(i))
        Else
            tbl.Cell(row, 3).Range.Text = "-"
        End If
        tbl.Cell(row, 4).Range.Text = arrCtx(i)
        row = row + 1
    Next i

    ' Nice column widths (percent)
    tbl.Columns(1).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(1).PreferredWidth = 6
    tbl.Columns(2).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(2).PreferredWidth = 39
    tbl.Columns(3).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(3).PreferredWidth = 8
    tbl.Columns(4).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(4).PreferredWidth = 47
End Sub

' ========================
' CONTEXT & SMALL HELPERS
' ========================
Private Function GetContextSnippet(doc As Document, ByVal startInDoc As Long, ByVal lengthInDoc As Long, ByVal wing As Long) As String
    Dim L As Long, R As Long
    L = MaxLng(0, startInDoc - wing)
    R = MinLng(doc.Content.End, startInDoc + lengthInDoc + wing)

    Dim leftR As Range, rightR As Range, hitR As Range
    Set hitR = doc.Range(startInDoc, startInDoc + lengthInDoc)
    Set leftR = doc.Range(L, startInDoc)
    Set rightR = doc.Range(startInDoc + lengthInDoc, R)

    GetContextSnippet = "…" & CleanOneLine(leftR.Text) & "[" & CleanOneLine(hitR.Text) & "]" & CleanOneLine(rightR.Text) & "…"
End Function

Private Function CleanOneLine(ByVal s As String) As String
    s = Replace$(s, vbCr, " ")
    s = Replace$(s, vbLf, " ")
    s = Replace$(s, Chr$(160), " ")
    Do While InStr(s, "  ") > 0
        s = Replace$(s, "  ", " ")
    Loop
    CleanOneLine = Trim$(s)
End Function

Private Function MinLng(a As Long, b As Long) As Long
    If a < b Then MinLng = a Else MinLng = b
End Function

Private Function MaxLng(a As Long, b As Long) As Long
    If a > b Then MaxLng = a Else MaxLng = b
End Function

Good luck

I want to tell my fellow lawyer coders about this new thing called Vibe Coding.

So, what is it? Vibe coding is an AI-powered development approach where a human communicates with an AI agent in natural language to generate functional code, focusing on high-level ideas and outcomes rather than intricate coding details. Popularised by OpenAI co-founder Andrej Karpathy in early 2025, this method allows users to describe desired app functionality and rely on the AI to handle implementation, accelerating development and making it accessible to those with limited programming experience.

Vibe coding allows you as a developer to concentrate on the creative aspects of app building, like user experience and functionality, instead of getting bogged down in technical specifics. Technical specifics always slowed down my coding and left me frustrated.

You essentially prompt your way to producing beautiful code.

Here is an example. Using your favourite AI product such as ChatGPT or Claude use the prompt

"create me a VBA macro that searches through a word document to find Capitalised terms." 

You will receive something like the visual basic code below, as your output from the prompt

Option Explicit

' =========================
' CAPITALISED TERMS FINDER
' =========================
' Scans the main body of the active document for Title-Case terms (1–5 words),
' skips first word of each sentence, deduplicates, counts, and reports:
' Term | Count | First Page | First Context.
'
' Run: FindCapitalisedTerms_Report
' -------------------------

Sub FindCapitalisedTerms_Report()
    Dim results As Object: Set results = CreateObject("Scripting.Dictionary")
    Dim stopList As Object: Set stopList = BuildStopList()
    
    ' ---- Regex setup (safe pattern) ----
    Dim rx As Object: Set rx = CreateObject("VBScript.RegExp")
    rx.Global = True
    rx.IgnoreCase = False
    rx.Multiline = True
    ' 1–5 Title-Case words; allow internal hyphen or apostrophe (straight or curly)
    ' NOTE: hyphen is first in the char class to avoid "invalid range" errors.
    rx.Pattern = "\b(?:[A-Z][a-z]{2,}(?:[-’'][A-Za-z]{2,})?)(?:\s+(?:[A-Z][a-z]{2,}(?:[-’'][A-Za-z]{2,})?)){0,4}\b"
    
    Dim doc As Document: Set doc = ActiveDocument
    
    ' Work only in the main story (avoids header/footer/textbox surprises)
    Dim bodyRng As Range
    Set bodyRng = doc.StoryRanges(wdMainTextStory)
    
    Dim s As Range, scanRng As Range
    Dim m As Object, matches As Object
    Dim term As String
    Dim startInDoc As Long, lengthInDoc As Long, pg As Long
    Dim ctx As String
    
    For Each s In bodyRng.Sentences
        If s.words.Count > 1 Then
            Set scanRng = s.Duplicate
            ' Skip the first word (most common false positive)
            On Error Resume Next
            scanRng.Start = s.words(2).Start
            If Err.Number <> 0 Then
                Err.Clear
                On Error GoTo 0
                GoTo NextSentence
            End If
            On Error GoTo 0
            
            ' Guard Execute with Test (avoids odd range errors)
            If rx.Test(scanRng.Text) Then
                Set matches = rx.Execute(scanRng.Text)
                For Each m In matches
                    term = Trim$(m.Value)
                    If ShouldKeepTerm(term, stopList) Then
                        startInDoc = scanRng.Start + m.FirstIndex
                        lengthInDoc = m.Length
                        
                        Dim hitRng As Range
                        Set hitRng = doc.Range(startInDoc, startInDoc + lengthInDoc)
                        
                        ' Page lookup can fail in unusual views/objects; guard it.
                        On Error Resume Next
                        pg = hitRng.Information(wdActiveEndAdjustedPageNumber)
                        If Err.Number <> 0 Then
                            Err.Clear: pg = 0
                        End If
                        On Error GoTo 0
                        
                        ctx = GetContextSnippet(doc, startInDoc, lengthInDoc, 45)
                        
                        If results.Exists(term) Then
                            Dim parts() As String
                            parts = Split(results(term), "|", 3)
                            parts(0) = CStr(CLng(parts(0)) + 1)
                            results(term) = Join(parts, "|")
                        Else
                            results.Add term, "1|" & CStr(pg) & "|" & ctx
                        End If
                    End If
                Next m
            End If
        End If
NextSentence:
    Next s
    
    RenderReport results
    MsgBox "Capitalised-terms report created.", vbInformation
End Sub

' -------------------------
' Filtering
' -------------------------
Private Function ShouldKeepTerm(ByVal term As String, ByVal stopList As Object) As Boolean
    Dim t As String: t = Trim$(term)
    
    ' Exclude ALL-CAPS tokens outright
    If t = UCase$(t) Then ShouldKeepTerm = False: Exit Function
    
    ' Single word?
    If InStr(t, " ") = 0 Then
        If Len(t) < 3 Then ShouldKeepTerm = False: Exit Function
        If stopList.Exists(t) Then ShouldKeepTerm = False: Exit Function
    Else
        ' For multi-word terms, keep if at least one word isn't on the stop list
        Dim words() As String, i As Long, ok As Boolean
        words = Split(t, " ")
        ok = False
        For i = LBound(words) To UBound(words)
            If Not stopList.Exists(words(i)) Then ok = True: Exit For
        Next i
        If Not ok Then ShouldKeepTerm = False: Exit Function
    End If
    
    ShouldKeepTerm = True
End Function

Private Function BuildStopList() As Object
    Dim d As Object: Set d = CreateObject("Scripting.Dictionary")
    Dim arr As Variant
    arr = Array( _
        "The", "A", "An", "And", "Or", "Of", "For", "To", "In", "On", "At", "By", _
        "I", "You", "We", "He", "She", "They", _
        "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday", _
        "January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December", _
        "Section", "Schedule", "Figure", "Table")
    Dim i As Long
    For i = LBound(arr) To UBound(arr): d(arr(i)) = True: Next i
    Set BuildStopList = d
End Function

' -------------------------
' Context & utilities
' -------------------------
Private Function GetContextSnippet(doc As Document, ByVal startInDoc As Long, ByVal lengthInDoc As Long, ByVal wing As Long) As String
    Dim L As Long, R As Long
    L = MaxLng(0, startInDoc - wing)
    R = MinLng(doc.Content.End, startInDoc + lengthInDoc + wing)
    
    Dim leftR As Range, rightR As Range, hitR As Range
    Set hitR = doc.Range(startInDoc, startInDoc + lengthInDoc)
    Set leftR = doc.Range(L, startInDoc)
    Set rightR = doc.Range(startInDoc + lengthInDoc, R)
    
    GetContextSnippet = "…" & CleanOneLine(leftR.Text) & "[" & _
                        CleanOneLine(hitR.Text) & "]" & _
                        CleanOneLine(rightR.Text) & "…"
End Function

Private Function CleanOneLine(ByVal s As String) As String
    s = Replace$(s, vbCr, " ")
    s = Replace$(s, vbLf, " ")
    s = Replace$(s, Chr$(160), " ")
    Do While InStr(s, "  ") > 0
        s = Replace$(s, "  ", " ")
    Loop
    CleanOneLine = Trim$(s)
End Function

Private Function MinLng(a As Long, b As Long) As Long
    If a < b Then MinLng = a Else MinLng = b
End Function

Private Function MaxLng(a As Long, b As Long) As Long
    If a > b Then MaxLng = a Else MaxLng = b
End Function

' -------------------------
' Report rendering
' -------------------------
Private Sub RenderReport(results As Object)
    Dim outDoc As Document: Set outDoc = Documents.Add
    outDoc.Activate
    
    Selection.Style = wdStyleHeading1
    Selection.TypeText "Capitalised Terms"
    Selection.TypeParagraph
    Selection.Style = wdStyleNormal
    Selection.TypeText "Generated: " & Format(Now, "yyyy-mm-dd hh:nn")
    Selection.TypeParagraph: Selection.TypeParagraph
    
    Dim n As Long: n = results.Count
    If n = 0 Then
        Selection.TypeText "No capitalised terms found (given current heuristics)."
        Exit Sub
    End If
    
    Dim tbl As Table
    Set tbl = outDoc.Tables.Add(Selection.Range, n + 1, 4)
    With tbl
        .Style = "Table Grid"
        .Cell(1, 1).Range.Text = "Term"
        .Cell(1, 2).Range.Text = "Count"
        .Cell(1, 3).Range.Text = "First Page"
        .Cell(1, 4).Range.Text = "First Context"
        .Rows(1).Range.Bold = True
    End With
    
    ' Alphabetical sort for a traditional, predictable report
    Dim keys() As String, i As Long, row As Long, k As Variant, parts() As String
    ReDim keys(0 To n - 1)
    i = 0
    For Each k In results.keys: keys(i) = CStr(k): i = i + 1: Next k
    QuickSort keys, LBound(keys), UBound(keys)
    
    row = 2
    For i = LBound(keys) To UBound(keys)
        parts = Split(results(keys(i)), "|", 3)
        tbl.Cell(row, 1).Range.Text = keys(i)
        tbl.Cell(row, 2).Range.Text = parts(0)
        tbl.Cell(row, 3).Range.Text = parts(1)
        tbl.Cell(row, 4).Range.Text = parts(2)
        row = row + 1
    Next i
    
    tbl.Columns(1).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(1).PreferredWidth = 25
    tbl.Columns(2).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(2).PreferredWidth = 10
    tbl.Columns(3).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(3).PreferredWidth = 10
    tbl.Columns(4).PreferredWidthType = wdPreferredWidthPercent: tbl.Columns(4).PreferredWidth = 55
End Sub

' Simple case-insensitive quicksort for string arrays
Private Sub QuickSort(arr() As String, ByVal first As Long, ByVal last As Long)
    Dim i As Long, j As Long, pivot As String, temp As String
    i = first: j = last
    pivot = arr((first + last) \ 2)
    Do While i <= j
        Do While StrComp(arr(i), pivot, vbTextCompare) < 0: i = i + 1: Loop
        Do While StrComp(arr(j), pivot, vbTextCompare) > 0: j = j - 1: Loop
        If i <= j Then
            temp = arr(i): arr(i) = arr(j): arr(j) = temp
            i = i + 1: j = j - 1
        End If
    Loop
    If first < j Then QuickSort arr, first, j
    If i < last Then QuickSort arr, i, last
End Sub




How to use

-Open your Word doc which includes the contract (which includes Defined Terms).

-Press Alt+F11 → Insert → Module → paste the code from the AI product you used.

-Close the editor and run View → Macros → FindCapitalisedTerms_Report. A new document appears with a table: Term | Count | First Page | First Context.
 with the 

• Amazing. This would have taken weeks without Vibe coding!