well you need to be more specific on how the communication works.
what does the software offer as hardware signals that do any kind of change on what that could be detected elctronicially?
There are a lot of possabilities to realise something like that.
One important thing is:
must the transmitter be switched ON withing 20 microseconds after the software starts transmitting? Or is there some kind of sending requests that start establishing a connection
and if it takes one or two seconds until your transmitter is switched on
still all data will be transmitted?
Another way would be to put a microcontroller inbetween your software and the transmitter.
If this can work reliably will depend on the baudrate that you are using and on the available RAM of the microcontroller to buffer the transmitted data.
In this way as soon as the microcontroller receives something start buffering the received data switch on the transmitter and if transmitter is ready to send forward the data to the transmitter.
So what baudrate is the transmitter working on?
Do you have some check mechanisms if the data was received correctly and in case of not correctly receiving it a new request for the same data is done?
Then a simple checking of the Tx-line does it go HIGH-LOW would be enough to detect transmission starts but some bytes might get lost which would be catched up by the checking and would initiate sending the same datapacket a second time.
like No 5 (the robot) would say "We need more inpuut"
best regards Stefan