The Artificial Intelligence methods employed in Intelligent Computer Assisted Language Learning (ICALL) in principle makes it possible to individually support language learners. Second Language Acquisition (SLA) research and language teaching practitioners agree on the relevance of target language input adapted to the learner level. However, little systematic research has explored individually adapting input and how it impacts learners. Building on previous findings on apparent alignment between the complexity of learner input and their output (Chen & Meurers, 2019), the purpose of this study is to investigate how different challenge levels of adaptive input impact learners’ written output . We developed an ICALL system implementing a Complex Input Primed Writing task that selects texts for individual learners and ran an experiment grouping learners into four classes: no, low, medium, or high challenge in relation to the individual learners’ writing complexity. The results show that learners generally were able to align to low- and medium-level challenges, producing more complex writings after receiving the adaptively challenging input, but less so for the high challenge group. The study demonstrates how an ICALL system used in a regular language learning context can support SLA research into adaptive input and complexity alignment.