{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RUIYZ3LSOEWOPZYN5W5GP5JXH6","short_pith_number":"pith:RUIYZ3LS","schema_version":"1.0","canonical_sha256":"8d118ced72712ce7e70dedba67f5373f9bda7fa4f570ee80c63e160c9c0f7cdd","source":{"kind":"arxiv","id":"2605.24785","version":1},"attestation_state":"computed","paper":{"title":"PANDO: Efficient Multimodal AI Agents via Online Skill Distillation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haotian Shen, Yidi Miao, Yubo Li, Yuxin Liu","submitted_at":"2026-05-24T00:07:25Z","abstract_excerpt":"Recent advances in multimodal web agents often rely on increased inference-time computation, including rollout search, verifier passes, offline skill discovery, and specialist model stacks. This raises a central question: can a web agent become more efficient as it accumulates experience, rather than more expensive? We first analyze trajectories from VisualWebArena and identify three recurring sources of inefficiency: repeat-action loops, hidden discovery costs, and low prompt-cache reuse. We then introduce PANDO, a single-rollout online skill-distillation framework that maintains a structured"},"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":"2605.24785","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-24T00:07:25Z","cross_cats_sorted":[],"title_canon_sha256":"6161d3d17d7d6f22ace16ec96cc2405831f51a2aec552ab74a935aa718b3fe16","abstract_canon_sha256":"8803a73375a407cf2692efab44cd73d752ed9c5bfba69ecfafabb4b1b78ac4c0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:58.011205Z","signature_b64":"5v1vOIZbg7T86hjtE6kO50DMLdE7b8CyoUVsw11YcN3i0dWrOl3l9a1mCwQDvhjILGXH3K6RsBMRgTVT74WpAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8d118ced72712ce7e70dedba67f5373f9bda7fa4f570ee80c63e160c9c0f7cdd","last_reissued_at":"2026-05-26T01:03:58.010430Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:58.010430Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PANDO: Efficient Multimodal AI Agents via Online Skill Distillation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haotian Shen, Yidi Miao, Yubo Li, Yuxin Liu","submitted_at":"2026-05-24T00:07:25Z","abstract_excerpt":"Recent advances in multimodal web agents often rely on increased inference-time computation, including rollout search, verifier passes, offline skill discovery, and specialist model stacks. This raises a central question: can a web agent become more efficient as it accumulates experience, rather than more expensive? We first analyze trajectories from VisualWebArena and identify three recurring sources of inefficiency: repeat-action loops, hidden discovery costs, and low prompt-cache reuse. We then introduce PANDO, a single-rollout online skill-distillation framework that maintains a structured"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24785","kind":"arxiv","version":1},"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/2605.24785/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":"2605.24785","created_at":"2026-05-26T01:03:58.010573+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24785v1","created_at":"2026-05-26T01:03:58.010573+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24785","created_at":"2026-05-26T01:03:58.010573+00:00"},{"alias_kind":"pith_short_12","alias_value":"RUIYZ3LSOEWO","created_at":"2026-05-26T01:03:58.010573+00:00"},{"alias_kind":"pith_short_16","alias_value":"RUIYZ3LSOEWOPZYN","created_at":"2026-05-26T01:03:58.010573+00:00"},{"alias_kind":"pith_short_8","alias_value":"RUIYZ3LS","created_at":"2026-05-26T01:03:58.010573+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6","json":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6.json","graph_json":"https://pith.science/api/pith-number/RUIYZ3LSOEWOPZYN5W5GP5JXH6/graph.json","events_json":"https://pith.science/api/pith-number/RUIYZ3LSOEWOPZYN5W5GP5JXH6/events.json","paper":"https://pith.science/paper/RUIYZ3LS"},"agent_actions":{"view_html":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6","download_json":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6.json","view_paper":"https://pith.science/paper/RUIYZ3LS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24785&json=true","fetch_graph":"https://pith.science/api/pith-number/RUIYZ3LSOEWOPZYN5W5GP5JXH6/graph.json","fetch_events":"https://pith.science/api/pith-number/RUIYZ3LSOEWOPZYN5W5GP5JXH6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6/action/storage_attestation","attest_author":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6/action/author_attestation","sign_citation":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6/action/citation_signature","submit_replication":"https://pith.science/pith/RUIYZ3LSOEWOPZYN5W5GP5JXH6/action/replication_record"}},"created_at":"2026-05-26T01:03:58.010573+00:00","updated_at":"2026-05-26T01:03:58.010573+00:00"}