{"id":32525,"date":"2026-05-08T18:37:13","date_gmt":"2026-05-08T18:37:13","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/32525\/"},"modified":"2026-05-08T18:37:13","modified_gmt":"2026-05-08T18:37:13","slug":"ai-data-exchange-2026-casey-mulligan-on-driving-continual-processes-improvements","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/32525\/","title":{"rendered":"AI & Data Exchange 2026: Casey Mulligan on driving continual processes improvements"},"content":{"rendered":"

NOTE: Casey Mulligan left the Small Business Administration and is chief economist and chief regulatory officer at the Department of Health and Human Services<\/a>.<\/p>\n

In the Office of Advocacy in the Small Business Administration, there was one person who was reviewing thousands of proposed federal rules to help determine their impact on small businesses.<\/p>\n

He was using a software program that the office had developed.<\/p>\n

The problem was this one employee was the only person who knew how to use that application, which is based on STATA, a statistical software tool developed by StataCorp nearly 40 years ago to manipulate and visualize data. And because the analysis under the Regulatory Flexibility Act is one of the top priorities for the Office of Advocacy, he couldn\u2019t keep up with the number of new rules, which caused a backlog.<\/p>\n

The Office of Advocacy is the watchdog for the Regulatory Flexibility Act. \u201cIt requires regulators to say something about the effect on small business and how many small businesses are you regulating with this rule. What industries would be affected? And things like that. There aren\u2019t a lot of people who have that data at their fingertips,\u201d said Casey Mulligan, the former chief counsel for advocacy at the SBA, during Federal News Network\u2019s AI & Data Exchange 2026<\/a>.<\/p>\n

\u201cNow he\u2019s one of our geniuses, but still every request in the whole government \u2014those 12,000 Biden administration rules. Now there are some exemptions and stuff, but there were still several thousand times this tool would be needed if the law was followed, and we only have one guy. So I said, \u2018Wait a second, you\u2019re one of my best guys, and you\u2019re bogged down in this bottleneck.\u2019 \u201d<\/p>\n

Mulligan said he wanted to make the tool available to the entire staff and relieve current and any future bottlenecks.<\/p>\n

\u201cI had a conversation with the artificial intelligence and asked, what can I do so that my whole staff can have this, so that every federal agency can have this, and eventually it convinced me that we probably should do this in HTML, which I\u2019m familiar with, and JavaScript, which I don\u2019t know anything about. The AI rebuilt our tool [in those languages],\u201d he said. \u201cIt\u2019s called Sextant, like in the navigation instrument that we used before we had GPS. It built the Sextant tool in HTML so that it could be on a web page. We have it now internally on a web page, so any one on the staff can use it.\u201d<\/p>\n

Leaning into the benefits of AI<\/p>\n

Mulligan said Sextant can generate a good part of the report that the Regulatory Flexibility Act requires.<\/p>\n

\u201cIt\u2019s not on our [public] web page yet. I can email it to people. There\u2019s a bit of bureaucracy to be allowed to put a tool like that on a government web page. So we\u2019re working on that part,\u201d he said. \u201cThey only took an afternoon to make the tool and it\u2019s taking about three or four months to get the approval to put on the web page.\u201d<\/p>\n

But once approved, the tool will be available to any agency or organization. As of May, at least a handful of agencies were using it internally.<\/p>\n

The Sextant tool is an example of how the Office of Advocacy is leaning into AI to improve processes. Mulligan said the initial success of the office in using AI tools is a credit to the staff\u2019s willingness to try new things.<\/p>\n

The Office of Advocacy is looking for opportunities to apply AI tools across more and more of its processes, especially with it being such a data-heavy organization.<\/p>\n

\u201cIt\u2019s great for data processing and writing code without syntax errors. So we\u2019ve been able to be a lot more productive that way. We\u2019re creating documents and reports,\u201d Mulligan said. \u201cComing from academia, it seems to me, even if it wasn\u2019t required, we should be sharing how are we using AI and publishing our various prompts and code. We developed a prompt suite that has maybe six of our most used prompts, and we built that into a dashboard. There\u2019s a fact-check button and that has a prompt behind it. We tend to use Perplexity. The tool has been perfected to avoid hallucination, cross check things, break whatever passage it is into sentences and deal with it one sentence at a time. That all happens in the background. What we get when we press the button is to see, \u2018Oh, here\u2019s a sentence. How accurate is it? What are the inaccuracies? What are the sources that we could see to look further?\u2019 \u201d<\/p>\n

The prompt suite is a set of generic instructions that gives the user an accuracy score between zero and 100. The large language model comes back with an accuracy score and then links back to the source to support the analysis, Mulligan explained.<\/p>\n

Improving interagency trust<\/p>\n

Mulligan said these cross checks happen every time someone writes a sentence in the prompt.<\/p>\n

\u201cThat saves us from having to type that prompt every time. All we do is present the passage, and it takes the same process \u2014 and that can be set up for auditing. You can have all that back and forth saved to files and things like that. So the idea is to reduce the friction between federal employees and the AI, so that we can use it more often, and faster and better,\u201d he said.<\/p>\n

At SBA, Mulligan shared that he wrote many memos and reports. \u201cThe fact-checking is helpful because I may refer to something as a rule, and it\u2019ll say, \u2018Well, that was actually just a policy statement.\u2019 Yeah, it was published in the Federal Register, and I had the page cited correctly, but I wasn\u2019t correct to call it a rule. I should call it a policy statement.\u201d<\/p>\n

Mulligan said the Office of Advocacy has been improving the interagency process of reviewing rules with the AI tools. Previously, if it sent the Environmental Protection Agency, for example, a memo or request with factual inaccuracies because the office didn\u2019t fully understand the topic, there would need to be back and forth to resolve any misunderstandings.<\/p>\n

But the large language models help the staff understand the topics better, which facilitates better communications with other agencies and helps get results faster, Mulligan shared.<\/p>\n

While at SBA, he also created an AI tool to help him keep track of meetings and priorities. That AI tool was developed it in Python, a coding language Mulligan doesn\u2019t know.<\/p>\n

\u201cThe tool takes meetings off my calendar and whatever notes I put there, usually just a few hand-picked notes in there, and it expands on those notes to create a more fulsome set of meeting notes. It presents them to me, asks me some questions and then I approve or edit the notes. That\u2019s a pretty efficient process for me, and it\u2019s also good exercise for me at the end of the week to remember what I promised people and things like that,\u201d he said. \u201cThen, the notes are saved in a database that is intelligently searchable, so I can ask it questions about when\u2019s the last time I met with this person or what are the highest priority asks that I had from my meetings in the last month.\u201d<\/p>\n

Discover more articles and videos now on the AI & Data Exchange event page<\/a>.<\/p>\n

Copyright
\n \u00a9\u00a02026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.\n <\/p>\n","protected":false},"excerpt":{"rendered":"NOTE: Casey Mulligan left the Small Business Administration and is chief economist and chief regulatory officer at the…\n","protected":false},"author":2,"featured_media":32526,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[24,20800,20789,25,20824,20825,20826,20827],"class_list":["post-32525","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","tag-ai","tag-ai-data-exchange","tag-ai-exchange-2026","tag-artificial-intelligence","tag-casey-mulligan","tag-office-of-advocacy","tag-sba","tag-small-business-administration"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/32525","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=32525"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/32525\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/32526"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=32525"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=32525"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=32525"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}