{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3VL6MYWTC765UAX5WANBX6H5BY","short_pith_number":"pith:3VL6MYWT","schema_version":"1.0","canonical_sha256":"dd57e662d317fdda02fdb01a1bf8fd0e292f1078ded82201d1f1b07cf38e0c7d","source":{"kind":"arxiv","id":"2508.16987","version":1},"attestation_state":"computed","paper":{"title":"WebSight: A Vision-First Architecture for Robust Web Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.AI","authors_text":"Asanshay Gupta, Tanvir Bhathal","submitted_at":"2025-08-23T11:02:59Z","abstract_excerpt":"We introduce WebSight, a vision-based autonomous web agent, designed to interact with web environments purely through visual perception, eliminating dependence on HTML or DOM-based inputs. Central to our approach we introduce our new model, WebSight-7B, a fine-tuned vision-language model optimized for UI element interaction, trained using LoRA on a web-focused subset of the Wave-UI-25K dataset. WebSight integrates this model into a modular multi-agent architecture, comprising planning, reasoning, vision-action, and verification agents, coordinated through an episodic memory mechanism.\n  WebSig"},"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":"2508.16987","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-08-23T11:02:59Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"9f12006fa792c66c0f569d82d87abc28ae909bf08194ab4808a2eb3aceb26acc","abstract_canon_sha256":"e7d210fe6bb1575e01f083e21a48832c5a557c0a38b5088c25b3ca859112c369"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:58:07.724146Z","signature_b64":"9lrqi+hEEqvpHL5wWLjJPnfGq2x1omrCLh773pgPWZ72eQPFOO9uabDo1aEOP+9XXhXqtO4jUBRLcspRZ7vgAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dd57e662d317fdda02fdb01a1bf8fd0e292f1078ded82201d1f1b07cf38e0c7d","last_reissued_at":"2026-07-05T11:58:07.723679Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:58:07.723679Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"WebSight: A Vision-First Architecture for Robust Web Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.AI","authors_text":"Asanshay Gupta, Tanvir Bhathal","submitted_at":"2025-08-23T11:02:59Z","abstract_excerpt":"We introduce WebSight, a vision-based autonomous web agent, designed to interact with web environments purely through visual perception, eliminating dependence on HTML or DOM-based inputs. Central to our approach we introduce our new model, WebSight-7B, a fine-tuned vision-language model optimized for UI element interaction, trained using LoRA on a web-focused subset of the Wave-UI-25K dataset. WebSight integrates this model into a modular multi-agent architecture, comprising planning, reasoning, vision-action, and verification agents, coordinated through an episodic memory mechanism.\n  WebSig"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.16987","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/2508.16987/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":"2508.16987","created_at":"2026-07-05T11:58:07.723736+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.16987v1","created_at":"2026-07-05T11:58:07.723736+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.16987","created_at":"2026-07-05T11:58:07.723736+00:00"},{"alias_kind":"pith_short_12","alias_value":"3VL6MYWTC765","created_at":"2026-07-05T11:58:07.723736+00:00"},{"alias_kind":"pith_short_16","alias_value":"3VL6MYWTC765UAX5","created_at":"2026-07-05T11:58:07.723736+00:00"},{"alias_kind":"pith_short_8","alias_value":"3VL6MYWT","created_at":"2026-07-05T11:58:07.723736+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2601.18842","citing_title":"GUIGuard-Bench: Toward a General Evaluation for Privacy-Preserving GUI Agents","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09571","citing_title":"Tuning Qwen2.5-VL to Improve Its Web Interaction Skills","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY","json":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY.json","graph_json":"https://pith.science/api/pith-number/3VL6MYWTC765UAX5WANBX6H5BY/graph.json","events_json":"https://pith.science/api/pith-number/3VL6MYWTC765UAX5WANBX6H5BY/events.json","paper":"https://pith.science/paper/3VL6MYWT"},"agent_actions":{"view_html":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY","download_json":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY.json","view_paper":"https://pith.science/paper/3VL6MYWT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.16987&json=true","fetch_graph":"https://pith.science/api/pith-number/3VL6MYWTC765UAX5WANBX6H5BY/graph.json","fetch_events":"https://pith.science/api/pith-number/3VL6MYWTC765UAX5WANBX6H5BY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY/action/storage_attestation","attest_author":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY/action/author_attestation","sign_citation":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY/action/citation_signature","submit_replication":"https://pith.science/pith/3VL6MYWTC765UAX5WANBX6H5BY/action/replication_record"}},"created_at":"2026-07-05T11:58:07.723736+00:00","updated_at":"2026-07-05T11:58:07.723736+00:00"}