Responsible AI as a scaling function<\/h2>\n\n\n\n
Our Office of Responsible AI creates and implements standardized RAI policies and processes, which speed up reviews and allow for consistency across products. The goal is to build responsible AI into the design and development of agentic AI systems, rather than treat it as something added at the end.<\/p>\n\n\n\n \u201cIf you start with responsible AI principles by design, privacy by design, and security by design, the impact in terms of resources is extremely low. And the benefit will be projected across the entire lifecycle of your solution.\u201d<\/p>\nAlberto Carrozzo, senior technical program manager, Microsoft Europe, the Middle East, and Africa (EMEA)<\/cite><\/blockquote>\n\n\n\n That approach is reflected in the work of Alberto Carrozzo and his team, who are experts at helping apply responsible AI practices across internal AI scenarios.<\/p>\n\n\n\n \u201cIf you start with responsible AI principles by design, privacy by design, and security by design, the impact in terms of resources is extremely low,\u201d says Carrozzo, a senior technical program manager and an RAI leader for Microsoft Europe, the Middle East, and Africa (EMEA). \u201cAnd the benefit will be projected across the entire lifecycle of your solution.\u201d<\/p>\n\n\n\n Our responsible AI standard<\/strong><\/strong><\/p>\n\n\n\n Learn about responsible AI at Microsoft, including tools, practices, and policies.<\/a><\/p>\n<\/div>\n<\/div>\n\n\n\n As our teams across Microsoft apply more and more AI tools in their work every day, distinct risk profiles appear. Supply chain agents use business-confidential information and supplier data. Legal systems analyze sensitive records and must make defensible decisions about their relevance. Customer support tools put AI directly into conversations with external customers.<\/p>\n\n\n\n Carrozzo\u2019s team has completed more than 2,000 internal-tool reviews so far, and they\u2019re using their learnings to help prioritize and process a pipeline of thousands more.<\/p>\n\n\n\n Patterns from earlier work can direct reviewers toward cases that warrant closer attention, which gives our team a way to handle that volume consistently and efficiently.<\/p>\n\n\n\n \u201cWhen you have thousands of agents, you need to use what you learned from the past,\u201d Carrozzo says. \u201cThe more we learn, the more we need to use this knowledge to follow the risk and address it as fast as possible.\u201d<\/p>\n\n\n\n Across our Customer Zero experience, several practices have emerged that can help teams with adoption and scaling work. The first may seem counterintuitive: Sometimes scaling AI means intentionally limiting what an AI system can do.<\/p>\n\n\n\n Our cloud supply chain organization\u2014which sources, delivers, maintains, and ultimately decommissions the hardware behind all our company services\u2014had grown its volume of business fivefold in recent years, while staffing remained flat. They needed to innovate.<\/p>\n\n\n\n The organization now has more than 100 enterprise-grade AI agents in production. Analysis that previously took several people days to assemble can now be automatically prepared and updated continuously, giving planners a real-time starting point to begin their work.<\/p>\n\n\n\n \u201cIf I build a planning agent, it shouldn’t be able to answer a question about the weather. It needs to be grounded in the fact that it was created to do a certain job.\u201d<\/p>\nJeevan Pathuri, vice president of software engineering, Cloud Supply Chain, Azure Hardware Systems and Infrastructure<\/cite><\/blockquote>\n\n\n\n As our teams put more agents into production, testing revealed a practical issue common with AI tools: An application sometimes allows through more of the model\u2019s general capability than its business purpose requires. At enterprise scale, a broader range of capabilities outside the agent’s intended scope can lead to uncertainty with its potential behavior, making decisions less predictable and outcomes harder to anticipate.<\/p>\n\n\n\n Our Cloud Supply Chain group uses OneRAI to register AI capabilities, document models and evaluations, track approvals, and test those boundaries as part of its responsible AI process.<\/p>\n\n\n\n \u201cWith RAI, I get the benefit of running the Red Team Agent to make sure a tool is not doing things that are outside of its boundary,\u201d says Jeevan Pathuri, a vice president of software engineering in Cloud Supply Chain, part of Azure Hardware Systems and Infrastructure at Microsoft. \u201cIf I build a planning agent, it shouldn\u2019t be able to answer a question about the weather. It needs to be grounded in the fact that it was created to do a certain job.\u201d<\/p>\n\n\n\n Defining what an agent should and shouldn\u2019t do up front reduces ambiguity, minimizes rework, and accelerates deployment by making system behavior more predictable.<\/p>\n\n\n\n A narrow scope also helps employees see what the system was built to handle and where their own judgment should be applied. AI-generated recommendations remain simply inputs for a decision\u2014the person using the agent still owns the outcome.<\/p>\n\n\n\n Litigation and regulatory investigations often require our legal department to review large volumes of data for relevance, privilege, personally identifiable information, security risks, and confidentiality concerns.<\/p>\n\n\n\n After search and machine-learning filters reduce the total volume of material, attorneys from our Corporate, External and Legal Affairs (CELA) department may still need to read tens of thousands of documents individually.<\/p>\n\n\n\n \u201cThe AI-generated reasoning needed to be reproducible and explainable. We had to make sure that the insights were verifiable\u2014that it was not a hallucination and was accurately tied to the document itself.\u201d<\/p>\nEJ Bastien, general manager of discovery programs, Microsoft Corporate, External and Legal Affairs (CELA)<\/cite><\/blockquote>\n\n\n\n Generative AI creates an opportunity to apply case-specific instructions across a large collection of documents, classify and summarize content, and preserve a written reason for each decision. But in highly sensitive legal work, getting an answer isn\u2019t enough. Attorneys need to understand why the AI reached its conclusion and be able to verify it against the underlying evidence.<\/p>\n\n\n\n \u201cThe AI-generated reasoning needs to be reproducible and explainable,\u201d says EJ Bastien, a general manager of Discovery Programs in Microsoft CELA. \u201cWe had to make sure that the insights were verifiable\u2014that it was not a hallucination and was accurately tied to the document itself. This is so that any time an attorney looks at the output, they have a written explanation for why a given document would be considered relevant or not.\u201d<\/p>\n\n\n\n CELA\u2019s agentic system was built in Azure using Azure OpenAI so that internal documents would not be exposed to a public model. The agent is restricted to a curated document set, rather than searching broadly. Attorneys can inspect the explanation, find the supporting material, and test whether the interpretation holds up before they make a legal decision.<\/p>\n\n\n\n Bastien describes the RAI operating principle as \u201ctrust but verify.\u201d Citations help users confirm what a source says and can reveal other repositories that may need to be preserved or collected. A strictly scoped corpus of documents and durable outputs support a review method that legal teams can explain.<\/p>\n\n\n\n In customer support, the challenges come in real time as our teams handle customer calls from nearly 200 countries in more than 60 languages. Some of these cases can be highly specialized, and only a limited number of support engineers have the expertise to handle them.<\/p>\n\n\n\n \u201cThere may be five experts in the world on a particular issue, and they all speak English,\u201d says Bill Fillmore, a senior technical program manager in Customer Service and Support. \u201cBut we now have an AI-powered translation platform that enables us to communicate with that customer in their chosen language.\u201d<\/p>\n\n\n\n However, putting AI directly into customer conversations creates responsible AI challenges. The system interacts with customers and customer data while technology and its security, privacy, and governance requirements continue to evolve. Addressing the RAI requirements only once, before deployment, isn\u2019t enough.<\/p>\n\n\n\n \u201cRAI is a priority for us. We make sure that the guardrails are there first, as far as security, privacy, and governance go.\u201d<\/p>\nJosh Wong, senior product manager, Customer Zero Engineering<\/cite><\/blockquote>\n\n\n\n Ensuring that customer support use cases like front-door problem solving and partially autonomous support are grounded in responsible AI principles has been a game changer for the team.<\/p>\n\n\n\n \u201cRAI is a priority for us,\u201d says Josh Wong, a senior product manager in Customer Zero Engineering. \u201cWe make sure that the guardrails are there first, as far as security, privacy, and governance go. Whenever there are updates to those things, we prioritize making sure we incorporate them into our AI solutions. That\u2019s non-negotiable at Microsoft.\u201d<\/p>\n\n\n\n That practice gives the team a way to keep the solution aligned with our requirements as both the technology and its use case evolve. Rather than treating responsible AI as a one-time approval, we continually incorporate updated requirements into the AI solution.<\/p>\n\n\n\n With the lightning pace of change that comes with agentic development, we\u2019re not standing still at Microsoft when it comes to responsible AI work.<\/p>\n\n\n\n Pathuri\u2019s Cloud Supply Chain group is progressing from individual agents to end-to-end supply chain workflows driven by AI. Bastien sees Microsoft\u2019s legal teams prototyping more case-specific AI tools while preserving the same obligations around evidence and human judgment. Wong\u2019s and Fillmore\u2019s teams are exploring AI for case routing, front-door problem solving, and autonomous support functions.<\/p>\n\n\n\n \u201cThese guardrails solve the challenge of scaling AI safely without slowing teams down. Our goal is not to block innovation, but to make safe AI adoption the default path.\u201d<\/p>\nAlberto Carrozzo, senior technical program manager, Microsoft Europe, the Middle East, and Africa (EMEA)<\/cite><\/blockquote>\n\n\n\n As AI takes on a larger role in the enterprise, the lessons from our Customer Zero experience become more important. Our teams cannot eliminate every risk when developing AI-powered solutions, but they can establish evidence, ownership, and feedback mechanisms that reinforce responsible AI principles to support stronger data governance without losing sight of the larger goals.<\/p>\n\n\n\n This is where responsible AI becomes a scaling function.<\/p>\n\n\n\n \u201cThings like mandatory risk assessments before deployment, permission-based data access, continuous telemetry and monitoring, and automated detection of security and responsible AI risks are key to effective AI deployment,\u201d Carrozzo says. \u201cThese guardrails address the challenge of scaling AI safely without slowing teams down. Our goal is not to block innovation, but to make safe AI adoption the default path.\u201d<\/p>\n\n\n\n Here are some lessons that can aid your own efforts to turn responsible AI into an engineering discipline that supports safe, regular deployment:<\/p>\n\n\n\n <\/p>\n","protected":false},"excerpt":{"rendered":" As we move from AI experimentation to enterprise-scale deployment, a new challenge is emerging: How do we govern thousands of AI agents, workflows, and copilots across Microsoft without slowing innovation? We’ve learned through our Customer Zero journey\u2014where we use our technology and services first and communicate our learnings back to our product groups\u2014that responsible AI […]<\/p>\n","protected":false},"author":209,"featured_media":25814,"comment_status":"closed","ping_status":"closed","sticky":true,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_hide_featured_on_single":false,"_show_featured_caption_on_single":true,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[906],"tags":[199,920,904,237,949,950,903],"coauthors":[841],"class_list":["post-25813","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-frontier-company","tag-ai","tag-ai-and-copilot","tag-customer-zero","tag-governance","tag-responsible-ai","tag-risk-management","tag-security","m-blog-post"],"yoast_head":"\n![\"A](\"https:\/\/cm-edgetun.pages.dev\/insidetrack\/blog\/uploads\/prod\/2026\/10\/Alberto-Carrozzo.png\")
<\/figure>\n\n\n\n\n
![\"\"](\"https:\/\/cm-edgetun.pages.dev\/insidetrack\/blog\/uploads\/prod\/2025\/07\/Learn-how-p-g.png\")
<\/figure>\n<\/div>\n\n\n\nSupply chain: Making agent behavior more predictable<\/h2>\n\n\n\n
![\"A](\"https:\/\/cm-edgetun.pages.dev\/insidetrack\/blog\/uploads\/prod\/2026\/10\/Jeevan-Pathuri.png\")
<\/figure>\n\n\n\n\n
Legal review: Making outputs verifiable<\/h2>\n\n\n\n
![\"A](\"https:\/\/cm-edgetun.pages.dev\/insidetrack\/blog\/uploads\/prod\/2026\/10\/EJ-Bastien.png\")
<\/figure>\n\n\n\n\n
Customer support: Keeping responsible AI current as technology evolves<\/h2>\n\n\n\n
![\"A](\"https:\/\/cm-edgetun.pages.dev\/insidetrack\/blog\/uploads\/prod\/2026\/10\/Josh-Wong.png\")
<\/figure>\n\n\n\n\n
Looking ahead to the Frontier future<\/h2>\n\n\n\n
\n
Key takeaways<\/h3>\n\n\n\n
\n
Try it out<\/h3>\n\n\n\n
\n
Related links<\/h3>\n\n\n\n
\n