{"id":8018,"date":"2026-05-26T18:02:03","date_gmt":"2026-05-26T22:02:03","guid":{"rendered":"https:\/\/engineering.jhu.edu\/cmts\/?p=8018"},"modified":"2026-05-27T09:15:54","modified_gmt":"2026-05-27T13:15:54","slug":"introducing-the-wse-ai-gateway","status":"publish","type":"post","link":"https:\/\/engineering.jhu.edu\/cmts\/introducing-the-wse-ai-gateway\/","title":{"rendered":"Introducing the WSE AI Gateway"},"content":{"rendered":"\n<!-- Paste this into a WordPress Custom HTML block -->\n<style>\n.wh-featured-image.wp-post-image {\n  display: none !important;\n}\n<\/style>\n<style>\n  .cmts-ai-gateway-post {\n    --text: #172033;\n    --muted: #53627a;\n    --blue: #174a7c;\n    --blue-dark: #0b2f56;\n    --blue-soft: #eaf3fb;\n    --border: #d7e4f0;\n    --accent: #2f80ed;\n    --bg: #ffffff;\n    max-width: 900px;\n    margin: 0 auto;\n    color: var(--text);\n    font-family: inherit;\n    line-height: 1.65;\n    font-size: 1.05rem;\n  }\n\n  .cmts-ai-gateway-post h1,\n  .cmts-ai-gateway-post h2 {\n    line-height: 1.18;\n    color: var(--blue-dark);\n    margin: 0 0 1rem;\n  }\n\n  .cmts-ai-gateway-post h1 {\n    font-size: clamp(2rem, 4vw, 3.25rem);\n    letter-spacing: -0.03em;\n    margin-bottom: 1.25rem;\n  }\n\n  .cmts-ai-gateway-post h2 {\n    font-size: clamp(1.45rem, 2.5vw, 2rem);\n    margin-top: 3rem;\n    padding-top: 1.5rem;\n    border-top: 1px solid var(--border);\n  }\n\n  .cmts-ai-gateway-post p {\n    margin: 0 0 1.25rem;\n  }\n\n  .cmts-ai-gateway-post a {\n    color: var(--blue);\n    font-weight: 700;\n    text-decoration-thickness: 0.08em;\n    text-underline-offset: 0.18em;\n  }\n\n  .cmts-ai-gateway-post .lede {\n    font-size: clamp(1.15rem, 2vw, 1.35rem);\n    color: var(--muted);\n    margin-bottom: 1.75rem;\n  }\n\n  .cmts-ai-gateway-post .callout {\n    border: 1px solid var(--border);\n    background: linear-gradient(135deg, var(--blue-soft), #ffffff);\n    border-radius: 18px;\n    padding: 1.4rem 1.5rem;\n    margin: 2rem 0;\n  }\n\n  .cmts-ai-gateway-post .callout strong {\n    color: var(--blue-dark);\n  }\n\n  .cmts-ai-gateway-post .thesis {\n    border-left: 5px solid var(--accent);\n    padding: 1rem 1.25rem;\n    margin: 2rem 0;\n    background: #f7fbff;\n    font-size: 1.15rem;\n  }\n\n  .cmts-ai-gateway-post .cta {\n    margin-top: 2.5rem;\n    padding: 1.5rem;\n    border-radius: 18px;\n    background: var(--blue-dark);\n    color: #ffffff;\n  }\n\n  .cmts-ai-gateway-post .cta a {\n    color: #ffffff;\n  }\n\n  @media (max-width: 640px) {\n    .cmts-ai-gateway-post {\n      font-size: 1rem;\n    }\n\n    .cmts-ai-gateway-post .callout,\n    .cmts-ai-gateway-post .cta {\n      padding: 1.15rem;\n      border-radius: 14px;\n    }\n  }\n<\/style>\n\n<article class=\"cmts-ai-gateway-post\">\n  <p>A year or two ago, many AI conversations began with a basic question: how can I use this technology? Today, they often begin with a prototype, a data model, a partially working script, or a workflow someone has already built with tools like ChatGPT, Codex, Claude Code, or Copilot.<\/p>\n\n  <p>That shift is visible in our own work here at CMTS. We now see a much higher volume of AI-related requests and project conversations that arrive with a high level of specificity. Many faculty, staff, and students have moved beyond asking for AI tools, and are using AI to help define the requirement, develop the interface, and test the idea.<\/p>\n\n  <p>Recently, for example, a program manager with the Office of Research came to us with a nearly complete research dashboard prototype. He was asking us to help him refine it and move it from prototype to production. These types of requests are more common now.<\/p>\n\n  <p class=\"thesis\">AI has lowered the cost of a minimum viable product.<\/p>\n\n <p>That creates an enormous opportunity for WSE. The people closest to a problem can now help build the first version of a solution. However, it also creates a new institutional risk. A useful prototype can quickly touch institutional data, source code, student work, research processes, or administrative systems. The work may begin locally, but the responsibility does not stay local.<\/p>\n\n  <p>The answer is giving capable people shared infrastructure so they can build responsibly.<\/p>\n\n  <div class=\"callout\">\n    <p>That is why CMTS is rolling out the <strong>WSE AI Gateway<\/strong>: a Johns Hopkins Engineering platform for programmatic AI access, unified billing, model choice, project-level controls, and support.<\/p>\n  <\/div>\n\n  <h2>The Challenge Is Orchestration<\/h2>\n\n  <p>AI is becoming institutional infrastructure because it is now embedded in how people draft, analyze, code, teach, prototype, and operate.<\/p>\n\n  <p>The external data points in the same direction. <a href=\"https:\/\/www.mckinsey.com\/capabilities\/quantumblack\/our-insights\/the-state-of-ai\">McKinsey&#8217;s 2025 State of AI survey<\/a> found that 88% of organizations now regularly use AI in at least one business function, but only about one-third have begun scaling AI across the enterprise, and only 39% report enterprise-level EBIT impact.<\/p>\n\n  <p><a href=\"https:\/\/www.microsoft.com\/en-us\/worklab\/work-trend-index\/ai-at-work-is-here-now-comes-the-hard-part\">Microsoft and LinkedIn&#8217;s 2024 Work Trend Index<\/a> found that 75% of global knowledge workers were already using AI at work, and 78% of AI users were bringing their own AI tools.<\/p>\n\n  <p>Together, those numbers describe the same shift we are seeing at WSE. AI capability has become widely available (and embedded inline) faster than many institutions have built the operating models to support it.<\/p>\n\n  <p class=\"thesis\">The institutional challenge is no longer simply access and adoption. It is orchestration.<\/p>\n\n  <h2>From Bottleneck to Backbone<\/h2>\n\n  <p>When people across an institution can build more on their own, the technology team&#8217;s role changes. We do not need to own every use case, write every script, or become the bottleneck for every experiment.<\/p>\n\n  <p>But we do need to provide the backbone: approved access, unified billing, usage visibility, security guardrails, budget controls, and a support path people can actually use. That is more important now than ever.<\/p>\n\n  <p>The hard part is increasingly the last mile. A dashboard, research assistant, course demo, or automation may start as a local prototype. Before it becomes a reliable service, it needs the right account structure, data boundaries, model access, cost center, monitoring, and support model.<\/p>\n\n  <p>The same pattern is showing up in security research. <a href=\"https:\/\/www.netskope.com\/netskope-threat-labs\/cloud-threat-report\/generative-ai-2025\">Netskope&#8217;s 2025 Generative AI report<\/a> found that the amount of data sent to generative-AI apps increased more than 30-fold over the prior year, that 72% of generative-AI use in the enterprise was shadow IT driven by personal accounts, and that source code accounted for nearly half of data-policy violations involving generative-AI apps.<\/p>\n\n  <p>In higher education, <a href=\"https:\/\/www.educause.edu\/content\/2026\/the-impact-of-ai-on-work-in-higher-education\">EDUCAUSE&#8217;s 2026 report on AI and work<\/a> found that 94% of respondents had used AI tools for work within the past six months, while only 54% were aware of policies or guidelines meant to guide that use and 56% had used tools not provided by their institutions.<\/p>\n\n  <p class=\"thesis\">Responsible AI has to be embedded in the systems people actually use.<\/p>\n\n  <h2>The Support Model Is Changing Too<\/h2>\n\n  <p>AI is also harder to support for the same reason it is harder to govern. The user base has expanded in every direction at once. We now support people who are deeply capable with AI tools and are bootstrapping homegrown solutions, people who are new to AI but being asked to use it, and a much larger middle of researchers, instructors, analysts, administrators, and developers using AI in field-specific ways.<\/p>\n\n  <p>That changes the support queue. A request might involve prompt design, API keys, model behavior, data handling, cost, code generated by an assistant, a research workflow, a course activity, or an integration with another tool. At the same time, the number of tools, providers, models, and account types keeps growing.<\/p>\n\n  <p>The Gateway is only a solution to a subset of the requests. But it gives us a common starting point.<\/p>\n\n  <h2>What the WSE AI Gateway Provides<\/h2>\n\n  <p>The WSE AI Gateway gives faculty and staff a single, institutionally supported path for API-based AI work. Users can create projects, generate API keys, manage access, choose from supported models, control spending, and integrate AI into code, data workflows, research tools, and course demonstrations.<\/p>\n\n  <p>It is important to be clear about what this is and is not.<\/p>\n\n  <p>The Gateway is not meant to replace tools such as HopGPT, ChatGPT Edu, Microsoft Copilot, or Azure-based services. Those tools remain important for chat, productivity, and enterprise workflows. The WSE AI Gateway fills a different gap: <strong>programmatic model access<\/strong> for people who need to build with AI through APIs.<\/p>\n\n  <p>On a practical level, it gives WSE unified billing through JHU cost centers, so teams don&#8217;t need to manage separate vendor accounts, personal keys, or individual agreements just to experiment responsibly.<\/p>\n\n  <div class=\"cta\">\n    <p>Faculty and staff can learn more and access the Gateway at <a href=\"https:\/\/gateway.engineering.jhu.edu\">gateway.engineering.jhu.edu<\/a>.<\/p>\n  <\/div>\n<\/article>\n\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n","protected":false},"excerpt":{"rendered":"<p>A year or two ago, many AI conversations began with a basic question: how can I use this technology? Today, they often begin with a prototype, a data model, a&nbsp;<a href=\"https:\/\/engineering.jhu.edu\/cmts\/introducing-the-wse-ai-gateway\/\">&hellip;<\/a><\/p>\n","protected":false},"author":7,"featured_media":8039,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"footnotes":""},"categories":[384],"tags":[388],"class_list":["post-8018","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology","tag-artificial-intelligence","odd"],"acf":[],"_links":{"self":[{"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/posts\/8018","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/comments?post=8018"}],"version-history":[{"count":10,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/posts\/8018\/revisions"}],"predecessor-version":[{"id":8056,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/posts\/8018\/revisions\/8056"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/media\/8039"}],"wp:attachment":[{"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/media?parent=8018"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/categories?post=8018"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/engineering.jhu.edu\/cmts\/wp-json\/wp\/v2\/tags?post=8018"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}