{"id":25777,"date":"2026-06-02T11:00:27","date_gmt":"2026-06-02T18:00:27","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/?p=25777"},"modified":"2026-06-02T12:51:26","modified_gmt":"2026-06-02T19:51:26","slug":"frontier-tuning-teaching-ai-to-work-the-way-you-do","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/frontier-tuning-teaching-ai-to-work-the-way-you-do\/","title":{"rendered":"Frontier Tuning: Teaching AI to work the way you do"},"content":{"rendered":"<p><span data-contrast=\"auto\">Today at Microsoft Build, we\u00a0introduced\u00a0<\/span><span data-contrast=\"auto\">Frontier Tuning,\u00a0<\/span><span data-contrast=\"auto\">a new\u00a0approach\u00a0to making AI work the way your business does<\/span><span data-contrast=\"auto\">\u00a0by applying reinforcement learning<\/span><span data-contrast=\"auto\">\u00a0inside your compliance boundary\u00a0<\/span><span data-contrast=\"auto\">with your own data, processes, and conventions<\/span><span data-contrast=\"auto\">. W<\/span><span data-contrast=\"auto\">e\u2019re\u00a0announcing<\/span><span data-contrast=\"auto\">\u00a0<\/span><span data-contrast=\"auto\">private preview<\/span><span data-contrast=\"auto\">,<\/span><span data-contrast=\"auto\">\u00a0available through Forward Deployed Engineers<\/span><span data-contrast=\"auto\">,<\/span><span data-contrast=\"auto\">\u00a0and\u00a0<\/span><span data-contrast=\"auto\">upcoming availability in<\/span><span data-contrast=\"auto\">\u00a0<a href=\"https:\/\/www.microsoft.com\/en-us\/microsoft-365-copilot\/microsoft-copilot-studio\/?msockid=270172c7bee465633c576585bfa96487\">Microsoft Copilot Studio<\/a> and <a href=\"https:\/\/azure.microsoft.com\/en-us\/products\/ai-foundry\/?msockid=270172c7bee465633c576585bfa96487\">Microsoft Foundry<\/a><\/span><span data-contrast=\"auto\">.<\/span><span data-contrast=\"auto\">\u00a0<\/span><\/p>\n<h2><b><span data-contrast=\"none\">Inside\u00a0Frontier\u00a0Tuning\u00a0<\/span><\/b><span data-ccp-props=\"{&quot;335559738&quot;:320,&quot;335559739&quot;:160}\">\u00a0<\/span><\/h2>\n<p><span data-contrast=\"none\">Frontier Tuning has three parts that work together: the environment where learning happens; the unique inputs you provide from your own business; and the tuned output models, skills, and harness that the system produces.<\/span><span data-ccp-props=\"{&quot;335557856&quot;:16777215,&quot;335559738&quot;:180,&quot;335559739&quot;:180}\">\u00a0<\/span><\/p>\n<ul>\n<li aria-setsize=\"-1\" data-leveltext=\"%1.\" data-font=\"Segoe UI\" data-listid=\"1\" data-list-defn-props=\"{&quot;335551671&quot;:1,&quot;335552541&quot;:0,&quot;335559683&quot;:0,&quot;335559684&quot;:-1,&quot;335559685&quot;:720,&quot;335559991&quot;:360,&quot;469769242&quot;:[65533,0,46],&quot;469777803&quot;:&quot;left&quot;,&quot;469777804&quot;:&quot;%1.&quot;,&quot;469777815&quot;:&quot;hybridMultilevel&quot;}\" data-aria-posinset=\"1\" data-aria-level=\"1\"><b><span data-contrast=\"none\">A continuously evolving environment.<\/span><\/b><span data-contrast=\"none\"> Tuning runs in a managed\u00a0Reinforcement\u00a0Learning\u00a0Environment (RLE) used both for post-training and inference.\u00a0During training, the system learns from real workflows, tool\u00a0usage, and eval signals without\u00a0affecting<\/span><b><span data-contrast=\"none\">\u00a0<\/span><\/b><span data-contrast=\"none\">production systems.\u00a0At\u00a0inference\u00a0it explores multiple\u00a0frontier and fine-tuned\u00a0models, from\u00a0Microsoft AI\u00a0and\u00a0OpenAI,\u00a0across turns to\u00a0find\u00a0stronger\u00a0candidate paths\u00a0before returning an answer.\u00a0The system improves continuously as it learns from each interaction.<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335557856&quot;:16777215,&quot;335559738&quot;:180,&quot;335559739&quot;:180,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<\/ul>\n<ul>\n<li aria-setsize=\"-1\" data-leveltext=\"%1.\" data-font=\"Segoe UI\" data-listid=\"1\" data-list-defn-props=\"{&quot;335551671&quot;:1,&quot;335552541&quot;:0,&quot;335559683&quot;:0,&quot;335559684&quot;:-1,&quot;335559685&quot;:720,&quot;335559991&quot;:360,&quot;469769242&quot;:[65533,0,46],&quot;469777803&quot;:&quot;left&quot;,&quot;469777804&quot;:&quot;%1.&quot;,&quot;469777815&quot;:&quot;hybridMultilevel&quot;}\" data-aria-posinset=\"2\" data-aria-level=\"1\"><b><span data-contrast=\"none\">Your company&#8217;s data, domain knowledge, and workflows, in one platform.<\/span><\/b><span data-contrast=\"none\">\u00a0You\u00a0bring\u00a0your\u00a0business data and\u00a0know-how\u00a0into the\u00a0RLE: content, processes, conventions, terminology, and\u00a0workflows\u00a0that collectively\u00a0define how\u00a0your\u00a0business\u00a0runs. The experience is built to be easy to use, with no need for a data science degree. With a simple, guided approach, teams can bring data in and start tuning right away,\u00a0enabling\u00a0more people within your organization to capture\u00a0the power of tuning.<\/span><span data-ccp-props=\"{&quot;335557856&quot;:16777215,&quot;335559738&quot;:180,&quot;335559739&quot;:180}\">\u00a0<\/span><\/li>\n<\/ul>\n<ul>\n<li aria-setsize=\"-1\" data-leveltext=\"%1.\" data-font=\"Segoe UI\" data-listid=\"1\" data-list-defn-props=\"{&quot;335551671&quot;:1,&quot;335552541&quot;:0,&quot;335559683&quot;:0,&quot;335559684&quot;:-1,&quot;335559685&quot;:720,&quot;335559991&quot;:360,&quot;469769242&quot;:[65533,0,46],&quot;469777803&quot;:&quot;left&quot;,&quot;469777804&quot;:&quot;%1.&quot;,&quot;469777815&quot;:&quot;hybridMultilevel&quot;}\" data-aria-posinset=\"3\" data-aria-level=\"1\"><b><span data-contrast=\"auto\">T<\/span><\/b><b><span data-contrast=\"none\">uned\u00a0models, skills, and\u00a0harness\u00a0that\u00a0stay\u00a0within\u00a0your compliance\u00a0boundary.\u00a0<\/span><\/b><span data-contrast=\"auto\">This system produces tuned models, embeddings, skills, orchestration logic, and a runtime harness. All of this runs on your data, with your controls, without leaving your compliance boundary. The models inherit your access controls, so only people who could see the underlying data can access models built from it. The tools are virtualized, so agents can improve without affecting production systems.<\/span><span data-ccp-props=\"{&quot;335557856&quot;:16777215,&quot;335559738&quot;:180,&quot;335559739&quot;:180}\">\u00a0<\/span><\/li>\n<\/ul>\n<p><span data-contrast=\"none\">Together, these three pieces form a loop that gets sharper with every\u00a0agent\u00a0interaction. As knowledge in\u00a0your\u00a0environment grows, the models and harness evolve with it, so\u00a0your\u00a0agents\u00a0keep getting better at the work\u00a0you\u00a0actually do.<\/span><span data-ccp-props=\"{&quot;335557856&quot;:16777215,&quot;335559738&quot;:180,&quot;335559739&quot;:180}\">\u00a0<\/span><\/p>\n<h2><b><span data-contrast=\"none\">Fits\u00a0the way you operate<\/span><\/b><span data-ccp-props=\"{&quot;335559738&quot;:320,&quot;335559739&quot;:160}\">\u00a0<\/span><\/h2>\n<p><a href=\"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-scaled.webp\"><img decoding=\"async\" class=\"alignnone size-large wp-image-25805\" src=\"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-1024x575.webp\" alt=\"Image of Fronteir Tuning showing\" width=\"1024\" height=\"575\" srcset=\"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-1024x575.webp 1024w, https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-300x168.webp 300w, https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-768x431.webp 768w, https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-1536x863.webp 1536w, https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-content\/uploads\/sites\/73\/2026\/06\/Frontier-Tuning-Final-1-2048x1150.webp 2048w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/a><\/p>\n<p><span data-contrast=\"auto\">Frontier Tuning\u00a0fits into how you already\u00a0build and\u00a0operate\u00a0agents.\u00a0Users\u00a0interact\u00a0with agents tuned\u00a0on\u00a0your company\u2019s\u00a0data\u00a0and\u00a0workflows. Makers\u00a0and developers\u00a0build and refine\u00a0these\u00a0agents in the tools\u00a0you\u2019re\u00a0already using, including\u00a0Microsoft\u00a0Copilot Studio and\u00a0Microsoft\u00a0Foundry.\u00a0<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559740&quot;:300}\">\u00a0<\/span><\/p>\n<p><span data-contrast=\"auto\">For example,\u00a0soon\u00a0within\u00a0<\/span><b><span data-contrast=\"auto\">Copilot Studio<\/span><\/b><span data-contrast=\"auto\">, you\u00a0will be able to access the\u00a0RLE and\u00a0use\u00a0data like transcripts, knowledge\u00a0bases\u00a0and Microsoft 365 artifacts\u00a0to improve the agents you\u00a0already rely on.\u00a0We\u2019re\u00a0also bringing capabilities to\u00a0<\/span><b><span data-contrast=\"auto\">Foundry<\/span><\/b><span data-contrast=\"auto\">\u00a0to allow\u00a0developers\u00a0to\u00a0tune\u00a0agents alongside the tools\u00a0you\u00a0already\u00a0use.\u00a0Here\u00a0too,\u00a0you\u2019ll\u00a0be\u00a0able to\u00a0set up an RLE, bring in your data, and tune\u00a0models,\u00a0including\u00a0Microsoft AI\u00a0models,\u00a0and\u00a0runtime behavior.\u00a0More details on\u00a0Foundry support\u00a0will come in the\u00a0coming months.\u00a0And today, Frontier Tuning is available in Private Preview through our\u00a0<\/span><b><span data-contrast=\"auto\">Forward Deployed Engineering<\/span><\/b><span data-contrast=\"auto\">\u00a0(FDE) team.\u00a0FDEs\u00a0can\u00a0partner with you\u00a0end to end: defining the scenario, setting eval\u00a0criteria, running the tuning\u00a0process\u00a0and\u00a0delivering the agent,\u00a0all within\u00a0your environment.\u00a0<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559740&quot;:300}\">\u00a0<\/span><\/p>\n<p><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559740&quot;:300}\">\u00a0<\/span><span data-contrast=\"auto\">Whether\u00a0you\u00a0build in Copilot Studio, develop in Foundry, or partner with an FDE, Frontier Tuning fits how your business already\u00a0operates.\u00a0<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559740&quot;:300}\">\u00a0<\/span><\/p>\n<h2><b><span data-contrast=\"none\">Frontier Tuning\u00a0in action<\/span><\/b><span data-ccp-props=\"{&quot;335559738&quot;:320,&quot;335559739&quot;:160}\">\u00a0<\/span><\/h2>\n<p><span data-contrast=\"auto\">Frontier Tuning\u00a0is\u00a0already\u00a0in the hands of customers.\u00a0We&#8217;ve\u00a0partnered\u00a0with a focused set of\u00a0organizations including\u00a0Land O&#8217;Lakes,\u00a0EY, Bristol Myers Squibb, Pearson, McKinsey, McCarthy T\u00e9trault, and the Josh Bersin Company.<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:300}\">\u00a0<\/span><\/p>\n<p><span data-contrast=\"auto\">\u201c<\/span><i><span data-contrast=\"auto\">Microsoft Frontier Tuning\u00a0enabled\u00a0us to generate significantly better Copilot outputs for Communication Coach. The results were more closely aligned with Pearson\u2019s learning science, giving learners clearer, more actionable feedback on how to strengthen workplace communication,\u201d said Gian Paolo Perrucci, Product &amp; Technology Officer,\u00a0<\/span><\/i><b><i><span data-contrast=\"auto\">Pearson.<\/span><\/i><\/b><\/p>\n<p><i><span data-contrast=\"auto\">&#8220;Microsoft Frontier Tuning is set to transform the tax practice across the global EY organization. By combining a tax-domain\u2013tuned reasoning LLM with our extensive enterprise knowledge and insights from our Tax Advisors, EY is elevating the delivery of tax services. Leveraging client context in Microsoft Work IQ and deep EY expertise, we are tuning an advisory agent within the RLE that will be deployed to 75,000 tax professionals globally in the coming period.&#8221; &#8211;\u00a0<\/span><\/i><i><span data-contrast=\"none\">Ben Ambrosino, EY Global Tax CTO,\u00a0<\/span><\/i><b><i><span data-contrast=\"none\">EY<\/span><\/i><\/b><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559685&quot;:720,&quot;335559739&quot;:160,&quot;335559740&quot;:300,&quot;469777462&quot;:[90],&quot;469777927&quot;:[0],&quot;469777928&quot;:[1]}\">\u00a0<\/span><\/p>\n<p><i><span data-contrast=\"auto\">&#8220;Microsoft\u00a0Frontier Tuning\u00a0has given us a powerful way to bring Galileo&#8217;s research-backed HR intelligence into bespoke agents inside the Copilot experience. It is one of the most compelling capabilities we have seen for putting deep domain expertise into the daily flow of work.&#8221;\u00a0&#8211; Josh Bersin, Founder and CEO,\u00a0<\/span><\/i><b><i><span data-contrast=\"auto\">The Josh Bersin Company<\/span><\/i><\/b><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559685&quot;:720,&quot;335559739&quot;:160,&quot;335559740&quot;:300,&quot;469777462&quot;:[90],&quot;469777927&quot;:[0],&quot;469777928&quot;:[1]}\">\u00a0<\/span><\/p>\n<p><i><span data-contrast=\"none\">&#8220;With Frontier Tuning, we\u2019re teaching the system how Microsoft HR works \u2013 capturing organizational knowledge in one connected environment that learns and improves with every use. We partnered with our product teams until the results were undeniable; successful task completion increased from 13% to 87%. Now we&#8217;re expanding to more HR workflows<\/span><\/i><i><span data-contrast=\"none\">.&#8221; &#8211;\u00a0<\/span><\/i><span data-contrast=\"none\">Nathalie\u00a0D&#8217;Hers,\u00a0<\/span><span data-contrast=\"none\">CVP Employee Experience<\/span><\/p>\n<p><span data-contrast=\"auto\">The pattern across these engagements is consistent: when you teach the system how your\u00a0organization\u00a0actually works,\u00a0you get much\u00a0higher fidelity output and more predictable execution.<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559737&quot;:720,&quot;335559738&quot;:320,&quot;335559739&quot;:160,&quot;335559740&quot;:280}\">\u00a0<\/span><\/p>\n<h2><b><span data-contrast=\"none\">Get started<\/span><\/b><span data-ccp-props=\"{&quot;335559738&quot;:320,&quot;335559739&quot;:160}\">\u00a0<\/span><\/h2>\n<p><span data-contrast=\"auto\">If you&#8217;re interested in Frontier Tuning, visit <\/span><a href=\"https:\/\/aka.ms\/frontiertuning\"><b><span data-contrast=\"auto\">aka.ms\/frontiertuning<\/span><\/b><\/a><span data-contrast=\"auto\">\u00a0to learn more. We will follow up with\u00a0additional\u00a0details.<\/span><span data-ccp-props=\"{}\">\u00a0<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>We&#8217;re announcing the private preview of Frontier Tuning, a new approach to making AI work the way your business does by applying reinforcement learning inside your compliance boundary with your own data, processes, and conventions.\u00a0<\/p>\n","protected":false},"author":214584,"featured_media":25805,"comment_status":"open","ping_status":"closed","sticky":true,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[],"class_list":["post-25777","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-microsoft-365-developer"],"acf":[],"blog_post_summary":"<p>We&#8217;re announcing the private preview of Frontier Tuning, a new approach to making AI work the way your business does by applying reinforcement learning inside your compliance boundary with your own data, processes, and conventions.\u00a0<\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/posts\/25777","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/users\/214584"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/comments?post=25777"}],"version-history":[{"count":2,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/posts\/25777\/revisions"}],"predecessor-version":[{"id":25810,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/posts\/25777\/revisions\/25810"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/media\/25805"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/media?parent=25777"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/categories?post=25777"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/microsoft365dev\/wp-json\/wp\/v2\/tags?post=25777"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}