The L13201 emitter is something I've looked at because it emits right on the sensitive wavelength for CO2 i.e. 4300 nm and it also has significant emissions at 3900 nm where CO2 has no effect: -
Vertical blue line added by me is the sensitivity at 3900 nm and the red line is about 4300 nm.
So, if you chose the Pyreos dual sensor of the type that has one channel sensitive to 3900 nm and the other channel at 4300 nm, you get a reference channel and a measurement channel all from one light source namely the L13201.
It's important to get the reference channel because that channel is unaffected by CO2 and can be used to stabilize the light output either by feedback or post compensation of the signals (in hardware but more likely software).
That reference channel is also subject to the same signal deteriorating factors that the CO2 channel suffers from such as contaminants such as water vapour or some other environmental effect. See below for the general idea: -
At 3900 nm there are no real signal perturbations due to any of the gases commonly associated with this part of the spectrum so it's "useful".
My problem now is that I can't understand, with this measure done, how
associate further measurement with the % of the gas.
The ratio of the measured signal to the reference signal is the approach taken by several CO2 sensor manufacturers such as City Technology and SGX Sensortech to name a few that I've come across. That ratio as it drops, is the measure of CO2 present in the sampled gas.