We propose a new method for fitting the full-shape of the Lyman-$alpha$ (Ly$alpha$) forest three-dimensional (3D) correlation function in order to measure the Alcock-Paczynski (AP) effect. Our method preserves the robustness of baryon acoustic oscillations (BAO) analyses, while also providing extra cosmological information from a broader range of scales. We compute idealized forecasts for the Dark Energy Spectroscopic Instrument (DESI) using the Ly$alpha$ auto-correlation and its cross-correlation with quasars, and show how this type of analysis improves cosmological constraints. The DESI Ly$alpha$ BAO analysis is expected to measure $H(z_mathrm{eff})r_mathrm{d}$ and $D_mathrm{M}(z_mathrm{eff})/r_mathrm{d}$ with a precision of $sim0.9%$ each, where $H$ is the Hubble parameter, $r_mathrm{d}$ is the comoving BAO scale, $D_mathrm{M}$ is the comoving angular diameter distance and the effective redshift of the measurement is $z_mathrm{eff}simeq2.3$. By fitting the AP parameter from the full shape of the two correlations, we show that we can obtain a precision of $sim0.5-0.6%$ on each of $H(z_mathrm{eff})r_mathrm{d}$ and $D_mathrm{M}(z_mathrm{eff})/r_mathrm{d}$. Furthermore, we show that a joint full-shape analysis of the Ly$alpha$ auto-correlation and its cross-correlation with quasars can measure the linear growth rate times the amplitude of matter fluctuations in spheres of $8;h^{-1}$Mpc, $fsigma_8(z_mathrm{eff})$. Such an analysis could provide the first ever measurement of $fsigma_8(z_mathrm{eff})$ at redshift $z_mathrm{eff}>2$. By combining this with the quasar auto-correlation in a joint analysis of the three high-redshift two-point correlation functions, we show that DESI could be able to measure $fsigma_8(z_mathrm{eff}simeq2.3)$ with a precision of $5-12%$, depending on the smallest scale fitted.