Hyaluronic acid (HA) is the natural glycosaminoglycan product of HA synthases with diverse cosmetic and medical applications. While process and/or metabolic engineering are utilized for industrial HA production, the potential of protein engineering has barely been realized. Herein, Knowledge‐gaining directed evolution (KnowVolution) was employed to generate an HA synthase variant from Pasteurella multocida (pmHAS) with improved chain length specificity and a two‐fold increase in mass‐based turnover number. Seven improved pmHAS variants out of 1392 generated by error‐prone PCR were identified, eight prospective positions were saturated and the most beneficial amino acid substitutions were recombined. After one round of KnowVolution, the longest HA polymer (up to 4.7 MDa), through an engineered pmHAS variant in a cell‐free system, was synthesized. Computational studies showed that the substitutions from the best variant (T40L, V59M and T104A) are distant from the glycosyltransferase sites and increased the flexibility of the N‐terminal region of pmHAS. Taken together, these findings suggest that the N‐terminus may be involved in HA synthesis and demonstrate the potential of protein engineering towards improved HA synthase activity.