<p>The Fenton-like reaction has attracted increasing attention in recent years due to the persistent release of organic pollutants from textile industries. Herein, Fe-doped Bi₁₂O₁₇Cl₂ (BiOCl) hierarchical catalysts were synthesised for photo-assisted Fenton degradation of textile dyes. Incorporation of iron species significantly enhanced degradation efficiency. For example, Fe-BiOCl-3 (k = 0.085&#xa0;min⁻¹) exhibited 2.5-fold higher performance than undoped BiOCl. The optimised Fe-BiOCl-3 catalyst, with moderate Fe loading (0.26 wt%), achieved degradation efficiencies of 99%, 74%, 99%, 97%, and 73.8% for rhodamine B, Bismarck brown, Congo red, indigo carmine, and orange G, respectively. The TOC removal of the representative rhodamine B dye in deionised water (72.5%) was higher than in stream water (42.8%) and industrial wastewater (53.6%), suggesting that water composition influences Fenton oxidation rates. The activity of Fe-BiOCl-3 in the absence of light (classical Fenton) or H₂O₂ (photocatalysis) was very slow. The enhanced catalytic performance arises from the synergy between Fenton oxidation and the photocatalytic process, where photogenerated electrons facilitate the conversion of Fe³⁺ to Fe²⁺, resulting in rapid H₂O₂ decomposition to generate ·OH radicals. Moreover, quenching experiments revealed that ·OH and ·O₂ were the main active species in the degradation process. Five machine learning models were trained to predict the degradation efficiency and the effects of process parameters in the Fenton-like process. Among ML models, the Random Forest model demonstrated the best performance, strong generalisation, and mechanistic consistency with the underlying photo-assisted Fenton process. The outcomes collectively demonstrated that the combined experimental–computational approach provided synergistic insights into the performance, mechanistic behaviour, and process optimisation of the Fe-BiOCl/H₂O₂/visible light system for practical wastewater treatment applications.</p> Graphical abstract <p></p>