Yeah I'm not them, but I think that's close.
What I would say is that our sensory data + mental models of that data are abstractions over whatever reality actually is, which I am guessing you agree with. On top of that I think something else that I am guessing you don't agree with: I think those abstractions are pretty arbitrary and hide important "implementation details" of reality that could turn out to have a big impact on what anything is or what we might do with it.
I am guessing the crux here is about how much those abstract models are constrained by the underlying reality. It seems like you believe that because sensory information/models converge in a repeatable and consistent way, that it must be close to accurate and complete, where by complete I don't mean we, eg, can "see" the whole EM spectrum, but that there being such a thing as an EM spectrum that we can only see part of gives us a reliable way to infer that our senses aren't missing any substantial chunks of conception space.
What I remain open to is that the abstractions are sort of self referential in that we're in a self constrained region of "potential sensory space," that we can sense and verify the things we can sense and verify, but that raw reality is deceptively huge and complex and a radically different sensory+model configuration might give us a qualitatively different picture of raw reality, even if it overlaps quite a bit with our current sensory bubble.
My main reason for being open to this is that it seems like QFT is an existence proof of our subjective awareness of reality being fundamentally askew from how raw reality actually works, and in fact we ended up narrowly optimized to experience things as we do and act as we do because of replication/evolution not because of correctness or how well our sensory abstraction carve reality at the actual joints.
I am guessing our disagreement is how much extra/unknown space there is, or how useful fundamentally different abstractions could be in interacting with raw reality stuff.