A framework for the computational design of elastic isotropic two-dimensional metamaterials is presented. It follows a rational or inverse design philosophy since the architected material is tailored to reach a target mechanical response [1]. Most studies rely on geometry-based inverse optimization techniques, neural networks, and generative adversarial networks, and its computational demand is yet prohibitive [2]. We take this challenge and present an analytical-based strategy. The proposed framework lies on a centro-symmetric heuristic molecule (HM) that constitutes the unit-cell of the periodic material [3,4]. The modelling of the HM is described through a rigid body spring model (RBSM) with linear formulation. Rigid shaped atoms interact through centred and non-centred spring-based bonds. Aiming at the additive manufacturing of the designed material, the theoretical response of the HM is compared against a continuum finite element (FE) model. A relationship between the micro and the desired macro elastic properties of the material is provided, such that the stiffness of bonds is found via an equivalent homogenized isotropic Cosserat continuum. Two architected elastic quasi-isotropic materials are designed, a ultra-soft auxetic material with a target Poisson's ratio of $\nu=-1.0$, and a standard soft material with a target Poisson's ratio of $\nu=0.2$. Under the assumptions of a single material and a non-assembly approach for the additive manufacturing step, the target elastic parameters are achievable within low deviations from an engineering standpoint: under 3\% regarding the target Poisson ratio, and 15\% regarding the anisotropy ratio. Such deviations are lower when atoms are significantly stiffer than the connections and when the connections have a residual bending stiffness in the atom-connection interface.