Artificial intelligence (AI) is increasingly shaping genome-editing design, supporting predictive modeling of TALEN- and CRISPR-based systems by integrating genomic sequence features, chromatin structural context, and experimental metadata. However, the rapid proliferation of machine learning (ML)-driven tools has created a fragmented landscape in which different models operate on distinct, incompatible scoring scales and rely on heterogeneous training datasets, thereby substantially limiting the systematic, reproducible benchmarking of these tools. It produces a clear unmet need for harmonization in this space. To address this gap, we propose an AI-driven meta-platform that integrates predictions from existing genome-editing design tools using a stacked ensemble learning strategy and meta-modeling. The framework harmonizes diverse datasets and incorporates continuously filtered feedback from experimentally validated editing outcomes, with appropriate safeguards to manage the challenges inherent to multi-source, community-contributed experimental data. Through this process, the system learns the context-dependent strengths and limitations of individual predictors across different cell types, delivery modalities, and nuclease variants. By doing so, it enables tool-agnostic benchmarking, mitigates model-specific biases, and provides confidence-aware prioritization of candidate guides. We argue that such a unifying platform can shift the field from isolated, model-specific predictions towards an integrated, knowledge-driven ecosystem. This is expected to enhance robustness, interpretability, and reproducibility while supporting responsible data governance through explainable AI and privacy-preserving learning.