[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-a-new-way-to-teach-ai-agents-your-skill-preferences":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":34,"tags":35,"sources":40,"feedback":44,"feedback_at":22,"cost_usd":44,"total_tokens":44},8352,"a-new-way-to-teach-ai-agents-your-skill-preferences","A New Way to Teach AI Agents Your Skill Preferences","A new architecture lets locally run AI agents track your habits and nudge which skill a remote LLM picks, without touching how it parses your request.","A new paper sketches a lightweight way for personal AI agents to learn which skill you actually want, without changing how the model behind them interprets your request.\n\nThe setup targets locally deployed personal agents that lean on a remote LLM to pick from a growing menu of external skills. Instead of asking that remote model to infer deeper meaning from a request, the researchers add a separate local layer that tracks statistical patterns in a user's past choices. That layer nudges the remote LLM's selection decision, while the model's parsing of what you actually asked for stays untouched. The pitch is personalization without running heavier, centralized learning on hardware that can't handle it.\n\nThat split matters because personal agents keep adding skills, and picking the wrong one is what makes an assistant feel clumsy rather than smart. Doing preference-tracking locally, and cheaply, means personalization without fine-tuning or retraining the remote model. The paper reports its method beats comparison approaches on regret and accuracy, but it does not name those baselines, publish numbers, or describe the benchmark used, so that comparison is the authors' own characterization rather than something readers can independently check.\n\nEvery agent framework promises smarter skill selection this year; the architecture split is the interesting bit here, not an unverified performance claim.","[\"ai agents\",\"llm\",\"personalization\",\"research\"]","2026-09-28T04:00:00.000Z","2026-09-29T02:17:19.450Z","2026-09-29T02:17:26.269Z","published",null,[24,30],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Pull the actual regret\u002Faccuracy figures and benchmark or testbed names from the paper (not just the abstract's unsupported superlatives) to back the central comparative claim, and fix the dek's claim that agents 'pick the right skill locally' since the source makes clear the remote LLM still performs the selection — only the preference statistics are local.","resolved",{"id":31,"reviewer":26,"round":32,"reason":33,"status":29},"editor-r2",2,"The dek now correctly attributes skill selection to the remote LLM, but the central comparative claim (lowest cumulative regret, highest test accuracy, significantly outperforming baselines) is still stated as fact in paragraph one and only undercut by a caveat in paragraph three — since the source has no benchmark names, numbers, or baseline comparisons at all, strip or heavily hedge that claim throughout rather than asserting then walking it back.","ai",[36,37,38,39],"ai agents","llm","personalization","research",[41],{"name":42,"url":43},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.05828",0,{"sections":46},[47,51,56,61,66,71,76,81,86,91,96,101,106,111],{"name":48,"slug":34,"count":49,"latest_published_at":50},"AI",4917,"2026-09-28T23:39:20.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Security","security",768,"2026-09-29T01:20:00.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Policy","policy",406,"2026-09-28T19:04:09.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Deals","deals",272,"2026-09-28T17:42:46.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Hardware","hardware",191,"2026-09-28T15:45:00.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Science","science",154,"2026-09-28T13:19:18.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Consumer Tech","consumer-tech",139,"2026-09-28T17:09:47.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Dev Tools","dev-tools",87,"2026-09-28T16:11:42.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Startups","startups",80,"2026-09-28T17:50:28.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"General","general",49,"2026-09-28T16:44:57.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":107,"slug":108,"count":109,"latest_published_at":110},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":112,"slug":113,"count":114,"latest_published_at":115},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]