Previous research has modelled the evolution of either knowledge creation or knowledge networks, but not their co-evolution. This work presents an agent-based model to cover this gap and challenge the intuition that both phenomena are mutually re-enforcing. The model consists on the rules of partner selection and the rules of knowledge creation by the agents. Agents in the knowledge network choose their partners depending on their previous collaboration history and on their attractiveness. Similarly, the amount of knowledge created by each agent depends on his number of partners and the knowledge he has created earlier. The simulations of the model show a wide variety of scenarios with different policy strategies suitable for each.