Natural metabolic pathways have been promoted as a route to green synthesis of biofuels and biochemicals. However, these processes can be difficult to engineer into chassis microorganisms, and, when successful, are often hampered by the limitations imposed by complex cellular metabolism such as toxicity of products and intermediates, slow growth rates, and maintaining cell viability. One potential solution to these problems is to perform the reactions in vitro with partially purified enzymes or cell lysates. Redox cofactor utilization is currently one of the major barriers to the realization of efficient and cost competitive cell-free biocatalysis, especially where multiple redox steps are concerned. The design of versatile, cofactor balanced modules for canonical metabolic pathways, such as glycolysis, is one route to overcoming such barriers. In this regard, the non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase (GapN) can be used as a shortcut into the glycolysis pathway by catalyzing the transformation of glyceraldehyde-3-phosphate into 3-phosphoglycerate. It is a crucial enzyme that can be involved in the regulation of ATP concentration in cell-free biocatalysis processes. However, GapN is strictly dependent on the NADP+ cofactor1, which prevents its use for NAD+ cofactor-dependent pathways or those based on the use of more stable and less expensive biomimetic cofactors. Therefore, we set up a computer-aided design framework to engineer GapN for enabling a NADH linked efficient cell-free glycolytic pathway with a net zero ATP usage. This rational design approach combines molecular dynamics simulations with Artificial Intelligence-based multistate computational design methods2-3 that allowed us to consider different conformational states encountered along the GapN enzyme catalytic cycle. In particular, the cofactor flip, characteristic of this enzyme family and occurring before product hydrolysis4, was taken into account to redesign the cofactor binding pocket for NAD+ utilization. While GapN exhibits only tiny trace activity with NAD+, an enhancement of more than 10,000-fold in this activity was achieved, corresponding to a recovery of more than 72 % of the activity of the wild-type enzyme on NADP+. The development of this engineered GapN variant with comparable kinetics to the WT enzyme on its native cofactor represents an important step towards establishing an efficient, robust, and versatile cell-free glycolysis module for biochemical syntheses that could have broad applications in the field of green chemistry. Moreover, the computer-aided engineering approach we set up here may also provide a template for other researchers wishing to conduct similar engineering campaigns.