[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-mnemosyne-puts-ai-workflow-actions-on-probation-before-committing":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},3189,"mnemosyne-puts-ai-workflow-actions-on-probation-before-committing","Mnemosyne Puts AI Workflow Actions on Probation Before Committing","A new open-source runtime treats every AI-generated workflow action as an untrusted proposal until it clears a deterministic constraint check.","AI agents can write plausible-looking steps that quietly break the systems they're supposed to fix.\n\nResearchers have published Mnemosyne, a runtime designed to sit between AI agents and the workflows they generate. The core idea, which they call Agentic Transaction Processing, is that any action an LLM, solver, or agent team proposes is held in suspense until a deterministic constraint set either admits or rejects it. The system maintains an append-only transition log and can perform localized repairs when something unexpected disrupts a running workflow — editing only the affected operations rather than scrapping and regenerating the whole plan. The team reports under 6% overhead for the projection-and-validation step, and their localized repair touched an order of magnitude fewer operations than a full global recompute would.\n\nThe safety argument here is unusually direct: correctness relative to the declared constraint set becomes independent of whether the proposing AI is competent, honest, or up to date. That is a meaningful shift from the current default, where most agentic systems trust generated output implicitly and discover problems only after something breaks. As AI agents get deployed in higher-stakes environments — logistics, finance, infrastructure — the gap between \"syntactically valid\" and \"actually safe to run\" becomes expensive.\n\nMnemosyne is open source, which lets practitioners audit the constraint model themselves rather than take the researchers' safety proofs on faith — a reasonable place to start the skepticism.","[\"ai\",\"agents\",\"open-source\",\"research\"]","2026-07-02T04:00:00.000Z","2026-07-02T04:12:10.306Z","2026-07-02T04:12:13.294Z","published",null,[],"ai",[24,26,27,28],"agents","open-source","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.00269",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"]