We study the question of enterprise models in enterprise software, where many transition rules are not hidden in the environment but stored explicitly as workflows, business rules, SLAs, schemas, and configuration records. We compare trained world models against discovery agents that inspect the live system at inference time to predict state changes.</p>\n<p>The main finding: learned world models can look strong in-distribution, but they are brittle under enterprise shift. Discovery agents are more robust because they recover the relevant dynamics from the active system itself.</p>\n<p>Would love to engage in researchers and people interested in world modeling and long-horizon planning under distribution shift.</p>\n","updatedAt":"2026-05-13T02:42:07.688Z","author":{"_id":"607f060442beb4da0f990182","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg","fullname":"Patrice Bechard","name":"patricebechard","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":10,"isUserFollowing":false,"primaryOrg":{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63d3095c2727d7888cbb54e2/Uv-Lx8PVGviqokfOyYlCN.png","fullname":"ServiceNow-AI","name":"ServiceNow-AI","type":"org","isHf":false,"plan":"enterprise"}}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9211843013763428},"editors":["patricebechard"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.12178","authors":[{"_id":"6a03de2386b054ce2fa40d4d","user":{"_id":"64a58f67aee3750f265d3f84","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64a58f67aee3750f265d3f84/LXLEQ-eamQExzQ-1o4bze.jpeg","isPro":false,"fullname":"Jishnu Nair","user":"jishnunair","type":"user","name":"jishnunair"},"name":"Jishnu Sethumadhavan Nair","status":"claimed_verified","statusLastChangedAt":"2026-05-13T07:50:57.337Z","hidden":false},{"_id":"6a03de2386b054ce2fa40d4e","user":{"_id":"607f060442beb4da0f990182","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg","isPro":false,"fullname":"Patrice Bechard","user":"patricebechard","type":"user","name":"patricebechard"},"name":"Patrice Bechard","status":"claimed_verified","statusLastChangedAt":"2026-05-13T07:51:00.860Z","hidden":false},{"_id":"6a03de2386b054ce2fa40d4f","name":"Rishabh Maheshwary","hidden":false},{"_id":"6a03de2386b054ce2fa40d50","user":{"_id":"603c6bf03249b99991dbcbd0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/603c6bf03249b99991dbcbd0/IZDS9PIg-9b8QleK08Zan.png","isPro":false,"fullname":"Surajit Dasgupta","user":"surajit","type":"user","name":"surajit"},"name":"Surajit Dasgupta","status":"claimed_verified","statusLastChangedAt":"2026-05-13T07:50:55.029Z","hidden":false},{"_id":"6a03de2386b054ce2fa40d51","name":"Sravan Ramachandran","hidden":false},{"_id":"6a03de2386b054ce2fa40d52","name":"Aakash Bhagat","hidden":false},{"_id":"6a03de2386b054ce2fa40d53","name":"Shruthan Radhakrishna","hidden":false},{"_id":"6a03de2386b054ce2fa40d54","name":"Pulkit Pattnaik","hidden":false},{"_id":"6a03de2386b054ce2fa40d55","name":"Johan Obando-Ceron","hidden":false},{"_id":"6a03de2386b054ce2fa40d56","name":"Shiva Krishna Reddy Malay","hidden":false},{"_id":"6a03de2386b054ce2fa40d57","name":"Sagar Davasam","hidden":false},{"_id":"6a03de2386b054ce2fa40d58","name":"Seganrasan Subramanian","hidden":false},{"_id":"6a03de2386b054ce2fa40d59","name":"Vipul Mittal","hidden":false},{"_id":"6a03de2386b054ce2fa40d5a","name":"Sridhar Krishna Nemala","hidden":false},{"_id":"6a03de2386b054ce2fa40d5b","name":"Christopher Pal","hidden":false},{"_id":"6a03de2386b054ce2fa40d5c","name":"Srinivas Sunkara","hidden":false},{"_id":"6a03de2386b054ce2fa40d5d","name":"Sai Rajeswar","hidden":false}],"publishedAt":"2026-05-12T00:00:00.000Z","submittedOnDailyAt":"2026-05-13T00:00:00.000Z","title":"Do Enterprise Systems Need Learned World Models? The Importance of Context to Infer Dynamics","submittedOnDailyBy":{"_id":"607f060442beb4da0f990182","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg","isPro":false,"fullname":"Patrice Bechard","user":"patricebechard","type":"user","name":"patricebechard"},"summary":"World models enable agents to anticipate the effects of their actions by internalizing environment dynamics. In enterprise systems, however, these dynamics are often defined by tenant-specific business logic that varies across deployments and evolves over time, making models trained on historical transitions brittle under deployment shift. We ask a question the world-models literature has not addressed: when the rules can be read at inference time, does an agent still need to learn them? We argue, and demonstrate empirically, that in settings where transition dynamics are configurable and readable, runtime discovery complements offline training by grounding predictions in the active system instance. We propose enterprise discovery agents, which recover relevant transition dynamics at runtime by reading the system's configuration rather than relying solely on internalized representations. We introduce CascadeBench, a reasoning-focused benchmark for enterprise cascade prediction that adopts the evaluation methodology of World of Workflows on diverse synthetic environments, and use it together with deployment-shift evaluation to show that offline-trained world models can perform well in-distribution but degrade as dynamics change, whereas discovery-based agents are more robust under shift by grounding their predictions in the current instance. Our findings suggest that, in configurable enterprise environments, agents should not rely solely on fixed internalized dynamics, but should incorporate mechanisms for discovering relevant transition logic at runtime.","upvotes":52,"discussionId":"6a03de2486b054ce2fa40d5e","ai_summary":"Enterprise discovery agents that read system configuration at runtime outperform traditional world models in configurable environments where dynamics change over time.","ai_keywords":["world models","enterprise systems","transition dynamics","deployment shift","runtime discovery","configuration reading","enterprise discovery agents","CascadeBench","world of workflows"],"organization":{"_id":"65f4df5de83b55da5d79fbb6","name":"ServiceNow-AI","fullname":"ServiceNow-AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63d3095c2727d7888cbb54e2/Uv-Lx8PVGviqokfOyYlCN.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"607f060442beb4da0f990182","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg","isPro":false,"fullname":"Patrice Bechard","user":"patricebechard","type":"user"},{"_id":"64a58f67aee3750f265d3f84","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64a58f67aee3750f265d3f84/LXLEQ-eamQExzQ-1o4bze.jpeg","isPro":false,"fullname":"Jishnu Nair","user":"jishnunair","type":"user"},{"_id":"69b9648f1cb897fdcb4ccbf7","avatarUrl":"/avatars/05645fec6b6bb1c00c88c6a3936cb790.svg","isPro":false,"fullname":"Sridhar Nemala","user":"sridharnemala","type":"user"},{"_id":"661708ddd5db35ab21925ff4","avatarUrl":"/avatars/55db3307ae86c636f4f475aa97c185e3.svg","isPro":false,"fullname":"Shiva Krishna Reddy Malay","user":"shivakrishnareddymalay","type":"user"},{"_id":"65036ffdb96045f918094fd6","avatarUrl":"/avatars/d57ceb8ced01b15f9321f6516d321a87.svg","isPro":false,"fullname":"Pulkit Pattnaik","user":"pulpat","type":"user"},{"_id":"64895ed451b69a8c82f493a4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64895ed451b69a8c82f493a4/0G3lDWoVWiFuFmPfXPKfQ.jpeg","isPro":false,"fullname":"Seganrasan Subramanian","user":"Seganrasan","type":"user"},{"_id":"65ea6558b985ae912c57b294","avatarUrl":"/avatars/54eeba9ce8cb5bf35edfa8fad76a813d.svg","isPro":false,"fullname":"Denis Akhiyarov","user":"dtanow","type":"user"},{"_id":"630a9d0665a977feb9a03718","avatarUrl":"/avatars/ecc971478acbf6550d889a03609f0070.svg","isPro":false,"fullname":"Saloni Mittal","user":"salmitta","type":"user"},{"_id":"603c6bf03249b99991dbcbd0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/603c6bf03249b99991dbcbd0/IZDS9PIg-9b8QleK08Zan.png","isPro":false,"fullname":"Surajit Dasgupta","user":"surajit","type":"user"},{"_id":"67d123bebfafeffab7130292","avatarUrl":"/avatars/6e2f0fe92a89afe3be64e517b5c7dcd6.svg","isPro":false,"fullname":"Arpit Katiyar","user":"ArpitKatiyar19","type":"user"},{"_id":"6669c547695c7c2fb9ade904","avatarUrl":"/avatars/2eacebcc7126309b3c97798170af440a.svg","isPro":false,"fullname":"Praison Selvaraj","user":"praison-selvaraj","type":"user"},{"_id":"64b935fc49bde5d94836202d","avatarUrl":"/avatars/4ede3a8cf5afc556070edc16c9262cbd.svg","isPro":false,"fullname":"Sourabh Surana","user":"sourabh89","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"65f4df5de83b55da5d79fbb6","name":"ServiceNow-AI","fullname":"ServiceNow-AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63d3095c2727d7888cbb54e2/Uv-Lx8PVGviqokfOyYlCN.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.12178.md"}">
Do Enterprise Systems Need Learned World Models? The Importance of Context to Infer Dynamics
Authors: ,
,
,
,
,
,
,
,
,
,
,
,
,
Abstract
Enterprise discovery agents that read system configuration at runtime outperform traditional world models in configurable environments where dynamics change over time.
AI-generated summary
World models enable agents to anticipate the effects of their actions by internalizing environment dynamics. In enterprise systems, however, these dynamics are often defined by tenant-specific business logic that varies across deployments and evolves over time, making models trained on historical transitions brittle under deployment shift. We ask a question the world-models literature has not addressed: when the rules can be read at inference time, does an agent still need to learn them? We argue, and demonstrate empirically, that in settings where transition dynamics are configurable and readable, runtime discovery complements offline training by grounding predictions in the active system instance. We propose enterprise discovery agents, which recover relevant transition dynamics at runtime by reading the system's configuration rather than relying solely on internalized representations. We introduce CascadeBench, a reasoning-focused benchmark for enterprise cascade prediction that adopts the evaluation methodology of World of Workflows on diverse synthetic environments, and use it together with deployment-shift evaluation to show that offline-trained world models can perform well in-distribution but degrade as dynamics change, whereas discovery-based agents are more robust under shift by grounding their predictions in the current instance. Our findings suggest that, in configurable enterprise environments, agents should not rely solely on fixed internalized dynamics, but should incorporate mechanisms for discovering relevant transition logic at runtime.
Community
We study the question of enterprise models in enterprise software, where many transition rules are not hidden in the environment but stored explicitly as workflows, business rules, SLAs, schemas, and configuration records. We compare trained world models against discovery agents that inspect the live system at inference time to predict state changes.
The main finding: learned world models can look strong in-distribution, but they are brittle under enterprise shift. Discovery agents are more robust because they recover the relevant dynamics from the active system itself.
Would love to engage in researchers and people interested in world modeling and long-horizon planning under distribution shift.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.12178 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.12178 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2605.12178 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.