The Evolvable Virtual Machine architecture (EVM) is a computing architecture based on the notion of distributed interactive asynchronously communicating agents. The EVM provides a massively decentralised and distributed asynchronous framework for experimenting with and studying properties of artificial evolutionary and open multi-agent systems. It can be used for multi-task learning and for automated program discovery. In this article, we discuss the template-based tagging mechanism and referrals extension to the EVM architecture. The new tagging and referrals mechanisms expands the information processing capabilities of the EVM architecture. Referrals provide more scalable automated search mechanisms that enable faster knowledge dissemination among distant EVM agents, or cells. In addition, referrals can be used as one of the mechanisms to combat parasitism among the EVM cells.