[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-memory-gets-a-user-profile-to-decide-what-to-recall":10,"sections":34},{"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":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},3217,"ai-memory-gets-a-user-profile-to-decide-what-to-recall","AI Memory Gets a User Profile to Decide What to Recall","A new retrieval framework called PPRO uses a running user profile to rank memories, outperforming both fixed-rule and training-free baselines on two benchmarks.","AI assistants that remember past conversations are only as useful as their ability to pull up the right memory at the right moment.\n\nResearchers have proposed Profile-guided Personalized Retrieval Optimization, or PPRO, a framework designed to make long-term conversational memory retrieval dependent on who is asking, not just what they asked. PPRO builds two memory banks — one episodic, one semantic — from past dialogue, then derives a user profile from those accumulated memories. That profile acts as a ranking prior, so retrieval weighs stable user traits, preferences, and relationships before surfacing results. A query rewriter is then trained using Group Relative Policy Optimization, with both retrieval quality and final answer quality feeding back as training signal, while the memory banks and answer model stay frozen.\n\nMost memory-augmented agents today rank recalled content by query similarity alone — a one-size-fits-all approach that ignores whether a fact matters to this particular user. PPRO's experiments on the LoCoMo and LongMemEval-S benchmarks show consistent gains over both training-free systems and trained baselines, with ablations confirming that the profile-guided ranking and the retrieval-oriented rewriting each carry independent weight.\n\nPersonalized retrieval is the part of the AI memory problem that labs tend to skip over in product announcements; this work suggests the gap is real and closeable, though the distance from a controlled benchmark to a messy, multi-year conversation history is considerable.","[\"ai\",\"machine learning\",\"personalization\",\"memory\"]","2026-07-02T04:00:00.000Z","2026-07-02T04:49:33.246Z","2026-07-02T04:49:36.139Z","published",null,[],"ai",[24,26,27,28],"machine learning","personalization","memory",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.00017",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,89,94,99],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":86,"slug":87,"count":83,"latest_published_at":88},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]