Mobile agents are a useful paradigm for network coding providing many advantages and disadvantages. Unfortunately, widespread adoption of mobile agents has been hampered by the disadvantages, which could be said to outweigh the advantages. There is a variety of ongoing work to address these issues, and this is discussed. Ultimately, genetic algorithms are selected as the most interesting potential avenue. Genetic algorithms have many potential benefits for mobile agents. The primary benefit is the potential for agents to become even more adaptive to situational changes in the environment and/or emergent security risks. There are secondary benefits such as the natural obfuscation of functions inherent to genetic algorithms. Pitfalls also exist, namely the difficulty of defining a satisfactory fitness function and the variable execution time of mobile agents arising from the fact that it exists on a network. DNAgents 1.0, an original application of genetic algorithms to mobile agents is implemented and discussed, and serves to highlight these difficulties. Modifications of traditional genetic algorithms are also discussed. Ultimately, a combination of genetic algorithms and artificial life is considered to be the most appropriate approach to mobile agents. This allows the consideration of agents to be organisms, and the network to be their environment. Towards this end, a novel framework called DNAgents 2.0 is designed and implemented. This framework allows the continual evolution of agents in a network without having a separate training and deployment phase. Parameters for this new framework were defined and explored. Lastly, an experiment similar to DNAgents 1.0 is performed for comparative purposes against DNAgents 1.0 and to prove the viability of this new framework.