[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-agents-that-split-searching-from-reading-beat-bigger-rivals":10,"sections":45},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":35,"tags":36,"sources":40,"feedback":44,"feedback_at":22,"cost_usd":44,"total_tokens":44},11282,"ai-agents-that-split-searching-from-reading-beat-bigger-rivals","AI agents that split searching from reading beat bigger rivals","A new research pipeline splits search and reading into separate AI agents, pushing document question-answering scores past bigger single-model systems.","A new multi-agent system called KnowForge beats much bigger all-in-one models at answering questions over huge, messy document collections - by refusing to let one model both search and read.\n\nKnowForge splits retrieval-augmented question answering into specialist jobs instead of handing one model every tool. A Locator agent only searches a hybrid knowledge-point index; a Deep Reader agent only reads the text, images and tables that search turns up; a rule-based judge forces a second search round when the first pass misses something; and a synthesizer checks the final answer's format. Running on a self-hosted Qwen3.6-35B-A3B model with just 3 billion parameters active at a time, it scored an F1 of 47.5 on the 2,441-question, 3,368-PDF M3DocVQA benchmark - 11.0 points above the published M3DocRAG system and 7.1 points above a one-shot retrieve-and-answer baseline on the same index and model. On MultiHop-RAG it hit 73.5 percent accuracy, above the best published RAG and GraphRAG scores (71.2), and on the single-document MMLongBench-Doc it lifted the same base model from 34.8 to 47.0, roughly matching GPT-4o reading raw page images (46.3).\n\nThe paper's ablations are the real argument: on M3DocVQA, the Deep Reader agent is worth 3.8 points; on MultiHop-RAG, 9.8 points; on MMLongBench-Doc, zero - a spread the authors say tracks how much each benchmark rewards careful reading over simply locating the right passage. Merge the two agents into one model with every tool and twice the rounds, and the score drops 1.9 F1, because the merged agent keeps searching instead of settling down to read.\n\nThat's a useful data point against the instinct to just give an agent more tools and more turns - sometimes the fix is a narrower job description, not a bigger one.","[\"ai\",\"rag\",\"document-qa\",\"llm-agents\"]","2026-10-12T04:00:00.000Z","2026-10-12T05:44:31.385Z","2026-10-12T05:44:35.553Z","published",null,[24,30],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"publisher-r1","publisher",1,"The Deep Reader's listed F1 gains (3.8, 9.8, and 0) are given in benchmark order M3DocVQA\u002FMultiHop-RAG\u002FMMLongBench-Doc, implying MultiHop-RAG gained 9.8 points, but the next sentence states MultiHop-RAG actually got zero additional F1 from the Deep Reader — the numbers and the text contradict each other.","resolved",{"id":31,"reviewer":32,"round":33,"reason":34,"status":29},"editor-r2","editor",2,"The source lists the Deep Reader's ablation gains (3.8, 9.8, 0) immediately after discussing the benchmarks in the order M3DocVQA\u002FMultiHop-RAG\u002FMMLongBench-Doc, implying MultiHop-RAG's gain is 9.8 — the draft still asserts MultiHop-RAG got 0 and just vaguely folds 3.8 and 9.8 into 'the two PDF-heavy benchmarks' without naming which benchmark got which figure, so the contradiction is hidden rather than fixed; explicitly attribute each of the three numbers to its benchmark per the source's stated o","ai",[35,37,38,39],"rag","document-qa","llm-agents",[41],{"name":42,"url":43},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.13128",0,{"sections":46},[47,50,55,60,65,70,75,80,85,90,95,100,105,110],{"name":48,"slug":35,"count":49,"latest_published_at":18},"AI",6966,{"name":51,"slug":52,"count":53,"latest_published_at":54},"Security","security",952,"2026-10-12T04:54:12.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Policy","policy",496,"2026-10-11T20:06:43.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Deals","deals",488,"2026-10-12T10:32:15.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Hardware","hardware",234,"2026-10-11T17:20:00.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",196,"2026-10-11T15:34:40.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Consumer Tech","consumer-tech",191,"2026-10-12T11:10:16.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Startups","startups",120,"2026-10-09T17:02:31.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"Software","software",115,"2026-10-11T07:34:45.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Dev Tools","dev-tools",111,"2026-10-12T11:22:24.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"General","general",70,"2026-10-11T14:15:00.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"Gaming","gaming",59,"2026-10-09T11:43:43.000Z",{"name":106,"slug":107,"count":108,"latest_published_at":109},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":111,"slug":112,"count":113,"latest_published_at":114},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]