According to the literature, some mineral filler is needed for the resin. In my understanding, this does not refer to aggregate (like sand), but a much finer mineral powder. After the filler is added, one can then decide if they are going to use aggregates. Filled geopolymer resin is analogous to pure portland cement. The filler is chemically significant.
Silica content and calcium content of the mineral filler both improve the strength of the resin ^1
I use silica flour and sometimes wollastonite. But I have recently learned that wollastonite is more chemically active than I had thought. Wollastonite is a naturally occurring mono calcium silicate (portland cement is tri calcium silicate which cures by forming a di calcium silicate gel). Apparently, in neutral water wollastonite significantly raises the pH of the solution, and probably the calcium can displace some of the sodium from the GP resin. Thus, a resin with the right amount of alkali which is then filled with wollastonite could have excess alkali in the cured sample, which would negatively affect the polymer (see discussion of Na-PSS). The same resin filled with silica flour may be just fine. This is a new discovery for me and I will pay attention to this effect in future tests.
^1 H. Xu, J. Deventer. The Geopolymerization of Natural Alumino-Silicates. Geopolymere ‘99