[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-llms-write-their-own-chemistry-rules-classifying-977-of-reactions":10,"sections":48},{"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":38,"tags":39,"sources":43,"feedback":47,"feedback_at":22,"cost_usd":47,"total_tokens":47},3203,"llms-write-their-own-chemistry-rules-classifying-977-of-reactions","LLMs Write Their Own Chemistry Rules, Classifying 97.7% of Reactions","A multi-agent LLM pipeline expanded a standard reaction taxonomy from 68 to 14,073 classes without human curation, matching a leading proprietary classifier.","An automated pipeline of large language models has taught itself to classify chemical reactions — and write the classification rules — across nearly 666,000 US patent reactions.\n\nResearchers built a multi-agent LLM framework that both labels reactions and generates the symbolic rules governing those labels, each rule verified against the full corpus before it sticks. Starting from a standard taxonomy of 68 reaction classes, the system expanded the ruleset to 14,073 classes with no human curation. A lightweight fingerprint classifier built on top of those rules then correctly classified 97.7% of reactions it had never seen before — matching a leading proprietary classifier while resolving chemistry at finer grain and extending to reaction types outside its training distribution.\n\nSynthesis planning software has long depended on fixed, hand-coded rulesets that break down at the edges of known chemistry. That constraint mattered because chemistry is long-tailed: rare reaction types are numerous in aggregate, and static libraries miss them. A system that writes and verifies its own rules on demand sidesteps that bottleneck, and the proprietary-parity accuracy figure suggests it does so without sacrificing reliability.\n\nThe real test is whether the pipeline holds up outside patent literature, where reaction reporting is unusually structured — messier lab notebooks and preprints will be a harder target.","[\"ai\",\"chemistry\",\"drug-discovery\",\"automation\"]","2026-07-02T04:00:00.000Z","2026-07-02T04:30:14.630Z","2026-07-02T04:30:17.701Z","published",null,[24,30,34],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"The body contains an unresolved editorial hedge ('the proprietary classifier it matched is unnamed, so that benchmark claim is hard to independently verify — worth keeping in mind') that reads as an inline critique fragment rather than a finished, publishable sentence; remove or convert it into neutral contextual framing before submission.","resolved",{"id":31,"reviewer":26,"round":32,"reason":33,"status":29},"editor-r2",2,"The open concern [editor-r1] is unresolved: the body still contains the inline editorial hedge about the unnamed proprietary classifier being hard to independently verify — remove or reframe that sentence as neutral context before resubmitting.",{"id":35,"reviewer":26,"round":36,"reason":37,"status":29},"editor-r3",3,"The body's closing sentence frames the next benchmark as a question the article raises but does not answer — acceptable — but the phrase 'matching a leading proprietary classifier' from the source is omitted entirely from the body, which is fine; however, the dek and body are otherwise clean and the editorial hedge has been removed; the sole remaining issue is that the body drops the 'matching a leading proprietary classifier' comparative claim without noting it, leaving the 97.7% figure without","ai",[38,40,41,42],"chemistry","drug-discovery","automation",[44],{"name":45,"url":46},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.01061",0,{"sections":49},[50,54,59,64,69,74,79,84,89,94,99,103,108,113],{"name":51,"slug":38,"count":52,"latest_published_at":53},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":55,"slug":56,"count":57,"latest_published_at":58},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":60,"slug":61,"count":62,"latest_published_at":63},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":65,"slug":66,"count":67,"latest_published_at":68},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":70,"slug":71,"count":72,"latest_published_at":73},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":75,"slug":76,"count":77,"latest_published_at":78},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":80,"slug":81,"count":82,"latest_published_at":83},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":85,"slug":86,"count":87,"latest_published_at":88},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":100,"slug":101,"count":97,"latest_published_at":102},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":104,"slug":105,"count":106,"latest_published_at":107},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":109,"slug":110,"count":111,"latest_published_at":112},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":114,"slug":115,"count":116,"latest_published_at":117},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]