We frequently talk about whether something is soluble in water as purely a function of temperature, because of course the equilibrium is a function of temperature (and slightly of pressure, or more than slightly, for gases). But equilibrium is an ideal state never reached; two different materials with the same equilibrium solubility might have very different dissolution rates, and in the absence of seed crystals, one might have a much higher energy barrier than the other to nucleate crystals. Moreover, the growth of crystals after nucleation is initially exponential, then slows down to quadratic, rather than the initially-linear growth you'd expect from the simple Boltzmann energy-barrier picture.
Similar comments pertain to other reactions: the Gibbs free energy determines the reaction's equilibrium, but not the reaction rate, which may be autocatalytic (like crystal growth, but with diffusion) and thus experience exponential growth.
Historically the humans haven't taken much advantage of this in material processing (?), other than in heat treatment of metals, where it's an unavoidable challenge. But most lab techniques involve running the relevant reactions fully to equilibrium over the time of seconds to hours, spanning about four orders of magnitude. Processes that don't happen to an appreciable degree over hours are considered unimportant; processes that happen in less than a second are considered immediate.
It occurs to me that modern electronics and microfluidics each offer us the opportunity to intervene reproducibly in such processes at nanosecond timescales, adding nine more orders of temporal magnitude to our arsenal. If we have two processes in a material mixture, one which runs to completion at a given temperature in 10 microseconds and the other in 10 milliseconds, we can interrupt the proceedings after 10 microseconds when the second process is only 0.1% complete, or perhaps 0.0001%. (Interrupt? For example, by diluting, chilling, or poisoning the interaction.) We commonly do this kind of thing over a longer timescale in cooking: boiling the carrots for five minutes may make them delightfully soft, while boiling them for an hour will make them unpalatably mushy.
The information needed to plan such processes is rarely available in the existing research literature, because workers in the field usually don’t care whether a particular material change takes a nanosecond, a millisecond, or a microsecond.
We can alternate between two processes, one which has a yield of 0.01% of a desired product (due to equilibrium, for example) and the other of which removes the product from it, at kilohertz or megahertz frequencies. Of course, there are many existing processes which work this way already without such alternation; the Pidgeon process, for example, produces magnesium through silicothermic reduction of magnesia (produced by calcining dolomite) despite an unpromising equilibrium, because the magnesium boils out of the reaction and the silicon is taken up by quicklime (also from the dolomite) to form larnite. But there are other processes that cannot be run in this way, for example because they involve ingredients that would have unwanted reactions with one another, or the desired equilibria require vastly different temperatures or pressures.