Every script needs a way to settle what its letters are, and most modern scripts have one in the character standard that encodes them. Ranjana does not. It has no Unicode allocation at all: a provisional slot exists on the roadmap and proposals have been put forward, but nothing is encoded, so there is no code point for ai or for any other Ranjana letter, and this atlas stores no character for them. What it stores instead is a drawing, a vector outline taken from an openly licensed Ranjana typeface, which is the same route already used here for other unencoded scripts. Two easier answers were available and both are wrong.
Storing a Devanagari or Newa character in place of a Ranjana one asserts that the letters are the same letter, which they are not; the Newa block encodes Prachalit, a different script used by the same community for the same languages. Drawing the letterforms by hand would be worse, since inventing shapes for a living liturgical script is a fabrication however carefully done. The outline route keeps the shape true to a real typeface and keeps the record honest about what is stored, which matters most for a script whose users are a small community still writing it. The absence of an encoding is not a historical curiosity but a present constraint. It means Ranjana cannot be searched, cannot be spell checked, cannot be reliably carried between one program and another, and cannot be read by the assistive software a visually impaired reader depends on. Whatever the eventual encoding decides, that decision will be made partly on the basis of charts and proposals rather than on the basis of the manuscripts, which is why the unsettled questions about the script's own inventory carry more weight than they otherwise would.