Out of curiosity, I put a k type thermocouple in the bean chute by the exhaust channel and used an Amprobe recording thermometer to record the exhaust temperature (named ExT) as it left the drum during a roast. The Amprobe can record and output a CSV file. I used the RW CSV download and superimposed the exhaust curve on the roast curve in a spreadshee. Image attached (hopefully). Maybe a bit crude, but some interresting results. I am presuming that cold air leakage from the front plate/drum gap isn’t affecting things too much. If I use the IBTS trace as actual bean temperature, then the ExT (after 2 minutes) is below the actual bean temperature. This tells me the fan setting (F3) is mostly causing cooling of the beans. The little blips in the ExT trace are from the tryer. Since it’s a pretty direct path from the tryer to the exhaust, I doubt it affects the general in-drum temperature much. I noticed that changes in power did not affect the curve much. Of course this behavior is probably highly dependent on batch size and ambient temperature. I’m curious what you all see in this and how you interpret it. Jim
With induction heating “drum temp” is a bit funny, because the drum is the heat source.
You could suck out all of the heat in the drum, and still have a hot drum.
So “drum temp” language might need some extra nuance when discussing the Bullet?
@gabyritaseek.qiAO mentioned a quote from Scott Rao that the Bullet is the least convective roaster he had ever used. And the aillio engineers claim the reason they did not include an air temp sensor is because they do not consider the airflow important enough, because the Aillio is a conductive roaster…. Which it may be just about the only conductive roaster, since most are roasting primarily with convective energy and conductive is secondary. Even Probat engineers have been noted as stating their machines are doing more convective than conductive heat transfer.
In personal use, I find I wish I had the heat build of F1, and the airflow of F5 or higher.
Adding heat to the airflow system helped, but there is too much plastic and the IBGT1 sensor all dislike this idea very very very much. 
ymmv.
And thank you for posting this graph, I had never seen this visually, but rather just take notes, so this is fascinating and I am learning so much!
This graph mirrors my experience. While I have measure different airflow locations and gotten different temps, so I do think the air temperature is likely higher inside the drum than what we measure, at least the measurements are consistent. And they do suggest with near certainty that the airflow is not actually getting to roasting temp targets, or that there are air leaks in the system at that point.