We summarize the findings of our paper $\tt \small arXiv:1701.08776$ $\tt \small[hep-th]$. We start by defining the quantum break-time. Once one understands a classical solution as expectation value of an underlying quantum state, it emerges as time-scale after which the true quantum evolution departs from the classical mean field evolution. We apply this idea to de Sitter space. Following earlier work, we construct a simple model of a spin-2 field, which for some time reproduces the de Sitter metric and simultaneously allows for its well-defined representation as coherent quantum state of gravitons. The mean occupation number $N$ of background gravitons turns out to be equal to the de Sitter horizon area in Planck units, while their frequency is given by the de Sitter Hubble parameter. In the semi-classical limit, we show that the model reproduces all semi-classical calculations in de Sitter, such as thermal Gibbons-Hawking radiation, all in the language of quantum $S$-matrix scatterings and decays of coherent state gravitons. Most importantly, this framework allows to capture the $1/N$-effects of back reaction to which the usual semi-classical treatment is blind. They violate the de Sitter symmetry and lead to a finite quantum break-time of the de Sitter state equal to the de Sitter radius times $N$. We also point out that the quantum-break time is inversely proportional to the number of particle species in the theory. Thus, the quantum break-time imposes the following consistency condition: Older and species-richer universes must have smaller cosmological constants. For the maximal, phenomenologically acceptable number of species, the observed cosmological constant would saturate this bound if our Universe were $10^{100}$ years old in its entire classical history.