There is no big mystery about this. When you speak of Brahmi, you probably
have the formal inscriptional style in mind, such as one sees on the rock
edicts of Asoka. Quite possibly early Brahmi when written on other
materials looked like that, but script classed as Brahmi was in use for a
long time, down to the end of the Kushana period (c200 - 250 CE). The
developments would have arisen quite naturally in the hands of scribes who
have written hundreds of thousands of pages throughout their lives. We can
see this with Roman letters on monuments and hand-written mss. You can try
this for yourself. Try writing fifty pages of a text in capital letters as
fast as you can and then compare the beginning with the end.
There is a very old Skt mss called the Spitzer Manuscript, dating from the
200 CE, which is written in a very fluent, sophiscated Brahmi hand. This
kind of script then developed further with the rise of the Gupta dynasty,
with the emergence of the so-called Imperial Gupta script, which was
presumably based on one of the currrent regional cursive Brahmi scripts in
fashion in the area of the Gupta court. With this script, we are more than
half-way to Devanagari, which gradually develops from Gupta script after the
Pala Dynasty. The role of regional developments, type of writing material
and implements (pens or stylus), would have played a part, which accounts
for the development of the rounded South Indian scripts from Brahmi. There
are specialist books which give pages and pages of script samples from each
phase and there you can see the natural transition from the primitive stone
inscriptions onwards.
And independently of Brahmi, there was the Aramaic-based Kharoshthi script
which was used in NW India and parts Central Asia. It is now becoming
recognized that documents (including the so-called Gandhari mss) in this
script were in wide use in that area for almost 400 years.