器官模型是研究组织更新的重要新工具。然而，表征底层分化动态仍具有挑战性。在这里，我们开发了TypeTracker，通过AI支持的细胞追踪和沿分支谱系树回溯终点命运的方式来识别细胞命运。最终迁移到绒毛的细胞在仍然深处于隐窝时就早早地选择了新类型，并带来了重要的后果：（i）分泌细胞在终端分裂前就已经做出选择，且分泌命运在姐妹细胞中对称出现。（ii）不同的分泌类型来自于不同的干细胞谱系，而不是一个全能的分泌祖细胞。（iii）承诺后增殖对分泌细胞和吸收细胞的比例产生了强烈的影响。（iv）空间格局形成是在承诺后通过类型相关的细胞重排而发生的。这种“承诺然后排序”的模型与传统的传送带模型形成对比，在后者中，细胞通过沿着隐窝-绒毛轴向上移动而分化，因此引发了关于底层承诺和排序机制的新问。

Organoids are a major new tool to study tissue renewal. However, characterizing the underlying differentiation dynamics remains challenging. Here, we developed TypeTracker, which identifies cell fates by AI-enabled cell tracking and propagating end point fates back along the branched lineage trees. Cells that ultimately migrate to the villus commit to their new type early, when still deep inside the crypt, with important consequences: (i) Secretory cells commit before terminal division, with secretory fates emerging symmetrically in sister cells. (ii) Different secretory types descend from distinct stem cell lineages rather than an omnipotent secretory progenitor. (iii) The ratio between secretory and absorptive cells is strongly affected by proliferation after commitment. (iv) Spatial patterning occurs after commitment through type-dependent cell rearrangements. This "commit-then-sort" model contrasts with the conventional conveyor belt picture, where cells differentiate by moving up the crypt-villus axis and hence raises new questions about the underlying commitment and sorting mechanisms.