ContextUnlike resilience, antifragility describes systems that get stronger rather than weaker under stress and chaos. Antifragile systems have the capacity to overcome stressors and come out stronger, whereas resilient systems are focused on their capacity to return to their previous state following a failure. As technology environments become increasingly complex, there is a great need for developing software systems that can benefit from failures while continuously improving. Most applications nowadays operate in cloud environments. Thus, with this increasing adoption of Cloud-Native Systems they require antifragility due to their distributed nature. ObjectiveThe paper proposes UNFRAGILE framework, which facilitates the transformation of existing systems into antifragile systems. The framework employs chaos engineering to introduce failures incrementally and assess the system's response under such perturbation and improves the quality of system response by removing fragilities and introducing adaptive fault tolerance strategies. MethodThe UNFRAGILE framework's feasibility has been validated by applying it to a cloud-native using a real-world architecture to enhance its antifragility towards long outbound service latencies. The empirical investigation of fragility is undertaken, and the results show how chaos affects application performance metrics and causes disturbances in them. To deal with chaotic network latency, an adaptation phase is put into effect. ResultsThe findings indicate that the steady stage's behaviour is like the antifragile stage's behaviour. This suggests that the system could self-stabilise during the chaos without the need to define a static configuration after determining from the context of the environment that the dependent system was experiencing difficulties. ConclusionOverall, this paper contributes to ongoing efforts to develop antifragile software capable of adapting to the rapidly changing complex environment. Overall, the research provides an operational framework for engineering software systems that learn and improve through exposure to failures rather than just surviving them.