{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BONUVSQV5LQBFAXO2BOHWKJMPB","short_pith_number":"pith:BONUVSQV","schema_version":"1.0","canonical_sha256":"0b9b4aca15eae01282eed05c7b292c7878f3534831b747b15e93161380903472","source":{"kind":"arxiv","id":"2504.14603","version":2},"attestation_state":"computed","paper":{"title":"UFO2: The Desktop AgentOS","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.HC","cs.OS"],"primary_cat":"cs.AI","authors_text":"Chao Du, Chaoyun Zhang, Chiming Ni, Dongmei Zhang, Fangkai Yang, He Huang, Jian-Guang Lou, Jian Mu, Jiaxu Qian, Liqun Li, Lu Wang, Minghua Ma, Pu Zhao, Qingwei Lin, Rujia Wang, Saravan Rajmohan, Shilin He, Si Qin, Suzhen Zheng, Yu Kang, Zhao Jiang","submitted_at":"2025-04-20T13:04:43Z","abstract_excerpt":"Recent Computer-Using Agents (CUAs), powered by multimodal large language models (LLMs), offer a promising direction for automating complex desktop workflows through natural language. However, most existing CUAs remain conceptual prototypes, hindered by shallow OS integration, fragile screenshot-based interaction, and disruptive execution.\n  We present UFO2, a multiagent AgentOS for Windows desktops that elevates CUAs into practical, system-level automation. UFO2 features a centralized HostAgent for task decomposition and coordination, alongside a collection of application-specialized AppAgent"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.14603","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-04-20T13:04:43Z","cross_cats_sorted":["cs.HC","cs.OS"],"title_canon_sha256":"ef71470d578b9df58c0f5f2d7c97a76c030588b450fb3214770b8b80c946ade3","abstract_canon_sha256":"d450fd778f0305ad54d18598fab2f9211874fe11e09af2c9648b5aff3761f0c9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:53:54.489256Z","signature_b64":"6IheyrCz+39lZaWGlW6DD9PhrGFcNwdiWVOzRLuavB2wuXrHmFtq8BSsdp+kcFomC15e/xUOw+CwK366luqJDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0b9b4aca15eae01282eed05c7b292c7878f3534831b747b15e93161380903472","last_reissued_at":"2026-07-05T10:53:54.488703Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:53:54.488703Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"UFO2: The Desktop AgentOS","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.HC","cs.OS"],"primary_cat":"cs.AI","authors_text":"Chao Du, Chaoyun Zhang, Chiming Ni, Dongmei Zhang, Fangkai Yang, He Huang, Jian-Guang Lou, Jian Mu, Jiaxu Qian, Liqun Li, Lu Wang, Minghua Ma, Pu Zhao, Qingwei Lin, Rujia Wang, Saravan Rajmohan, Shilin He, Si Qin, Suzhen Zheng, Yu Kang, Zhao Jiang","submitted_at":"2025-04-20T13:04:43Z","abstract_excerpt":"Recent Computer-Using Agents (CUAs), powered by multimodal large language models (LLMs), offer a promising direction for automating complex desktop workflows through natural language. However, most existing CUAs remain conceptual prototypes, hindered by shallow OS integration, fragile screenshot-based interaction, and disruptive execution.\n  We present UFO2, a multiagent AgentOS for Windows desktops that elevates CUAs into practical, system-level automation. UFO2 features a centralized HostAgent for task decomposition and coordination, alongside a collection of application-specialized AppAgent"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.14603","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2504.14603/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2504.14603","created_at":"2026-07-05T10:53:54.488776+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.14603v2","created_at":"2026-07-05T10:53:54.488776+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.14603","created_at":"2026-07-05T10:53:54.488776+00:00"},{"alias_kind":"pith_short_12","alias_value":"BONUVSQV5LQB","created_at":"2026-07-05T10:53:54.488776+00:00"},{"alias_kind":"pith_short_16","alias_value":"BONUVSQV5LQBFAXO","created_at":"2026-07-05T10:53:54.488776+00:00"},{"alias_kind":"pith_short_8","alias_value":"BONUVSQV","created_at":"2026-07-05T10:53:54.488776+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01936","citing_title":"What to Format and How: A Benchmark and Workflow Approach for Document Formatting","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00288","citing_title":"Model-Native Computing Architecture: Envisioning Future System Architecture Through the Lens of Computer Architecture","ref_index":172,"is_internal_anchor":false},{"citing_arxiv_id":"2509.06477","citing_title":"MAS-Bench: A Unified Benchmark for Shortcut-Augmented Hybrid Mobile GUI Agents","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2509.07553","citing_title":"VeriOS: Query-Driven Proactive Human-Agent-GUI Interaction for Trustworthy OS Agents","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2509.21982","citing_title":"RISK: A Framework for GUI Agents in E-commerce Risk Management","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2510.24168","citing_title":"MGA: Memory-Driven GUI Agent for Observation-Centric Interaction","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27996","citing_title":"Exploring LLM Agent Designs and Interaction Modalities for Scientific Visualization","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27996","citing_title":"Exploring LLM Agent Designs and Interaction Modalities for Scientific Visualization","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14113","citing_title":"UI-Zoomer: Uncertainty-Driven Adaptive Zoom-In for GUI Grounding","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13531","citing_title":"RiskWebWorld: A Realistic Interactive Benchmark for GUI Agents in E-commerce Risk Management","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15671","citing_title":"Long-Term Memory for VLA-based Agents in Open-World Task Execution","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB","json":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB.json","graph_json":"https://pith.science/api/pith-number/BONUVSQV5LQBFAXO2BOHWKJMPB/graph.json","events_json":"https://pith.science/api/pith-number/BONUVSQV5LQBFAXO2BOHWKJMPB/events.json","paper":"https://pith.science/paper/BONUVSQV"},"agent_actions":{"view_html":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB","download_json":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB.json","view_paper":"https://pith.science/paper/BONUVSQV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.14603&json=true","fetch_graph":"https://pith.science/api/pith-number/BONUVSQV5LQBFAXO2BOHWKJMPB/graph.json","fetch_events":"https://pith.science/api/pith-number/BONUVSQV5LQBFAXO2BOHWKJMPB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB/action/storage_attestation","attest_author":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB/action/author_attestation","sign_citation":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB/action/citation_signature","submit_replication":"https://pith.science/pith/BONUVSQV5LQBFAXO2BOHWKJMPB/action/replication_record"}},"created_at":"2026-07-05T10:53:54.488776+00:00","updated_at":"2026-07-05T10:53:54.488776+00:00"}