UChicago Brings Together 81 Employees to Create with Claude Code

The University of Chicago hosted a hackathon in March 2026, bringing together 81 members of its IT staff around Claude Code. Divided into 17 teams, they produced more than a dozen functional prototypes. Open to non-developers, this experience notably led to the creation of a tool used by human resources.
Human Resources Uses a Tool Developed from the Hackathon
The projects primarily focused on daily work, and some continue to be in use. Allison Kallo, the instructional design manager at UChicago, notes that the HR prototype provided the foundation for a tool currently employed in that department. Its function: to help HR and recruitment managers prepare consistent and compliant job descriptions more quickly.
The software categorizes the role, finds its match in the official catalog along with its code, and then prepares a first draft using validated templates and formulations. The recipient can edit it before distribution, without starting from a blank page. Another project generates synthetic data based on what a developer describes about their project and needs; the resulting datasets intentionally include irregularities and edge cases.
Another group designed a Chrome extension focused on languages: on any web page, it substitutes foreign terms for selected words, allowing the reader to look up meanings through context. The institution also created a monthly AI user group open to the entire campus, where participants present their creations. Valerie Archambeau, director of communication technology services at UChicago, reports that messaging exchanges regularly include requests for invites to try similar projects.
Personal Data Was Excluded from the Projects
For this event welcoming dozens of individuals without a developer profile, restrictions were defined in advance. No access was granted to production systems or testing environments. Any structured data had to be sanitized beforehand, and information regarding students, health, or individuals was excluded.
Tom Kompare, head of application development at UChicago, specifies that participants who brought their own data had to make it fictional. According to Valerie Archambeau, contracts with vendors also stipulate that university data remains within the institution's environment and is not used to train their tools. She recommends that interested institutions quickly involve their security specialists, as their policies exhibit varying degrees of maturity.
Accessibility was also part of the guidelines. The head of the accessibility group came to raise awareness among teams about considering it during construction. An additional prompt asked them to request AI compliance with WCAG 2.1 AA, specifically detailing contrast, alternative text, and keyboard navigation requirements.
UChicago Aimed to Go Beyond Developer Training
The university's deployment began in 2024, with Zoom AI Companion, Microsoft Copilot, and Phoenix AI, an internal instance of ChatGPT. Rules, training, and a resource site accompanied these tools. However, the adoption remained below the expectations of management, notes Valerie Archambeau.
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisissez votre rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
With several IT colleagues, she preferred a creation event over another series of courses. Initially intended to introduce developers to programming assistants, the project was later expanded to include other IT staff beyond the central service. For Valerie Archambeau, the goal was to address difficulties encountered at work, simplify procedures, or make otherwise lengthy tasks feasible, rather than just focusing on learning code.
Allison Kallo explains that management wanted a reassuring and playful space to try programming. Registration was open and voluntary. She describes participants who came out of curiosity, eager to understand the possibilities of technology and how to use it effectively.
Teams Mixed Professions and Skill Levels
Before the workdays, the IT services offered advice and meetings showcasing AI capabilities while gathering issues to address. Representatives from Amazon Web Services answered questions, and a temporary enterprise license was provided for the event. Astrid Fingerhut, head of identity and access management solutions and operations at UChicago, also mentioned individual office hours aimed at reassuring those who feared they wouldn’t understand.
Tom Kompare relied on a team already familiar with the tool and its applications to professional processes. He prepared a guide covering both its use and the progression of a software project: needs expression, design, implementation, testing, and then presentation. Future participants had proposed their ideas and indicated their skill level during registration, ranging from no experience with AI to daily use.
Groups were formed before the hackathon began. According to Tom Kompare, their composition aimed to blend AI experiences and professional skills, hoping to gather individuals interested in the same project. To expedite this initial distribution, he had the AI generate suggestions, which he then reviewed personally.
Mentors Supported Prototype Development
Over two days, participants worked either on-site or remotely. Armed with the guidelines provided earlier, the teams began designing and building their tools. Mentors could step in if there were any roadblocks: Tom Kompare, Arjun Lagisetty, UChicago's IT manager, along with several AWS members.
According to Tom Kompare, many groups had already made progress on needs and design by the end of the first day and were nearing completion. The following morning was reserved for final developments, testing, and preparation for presentations. Each team then presented their project in the afternoon using three slides.
The Next Edition Will Eliminate Remote Participation
The organizers are preparing a new edition for early next year. Participants found it more challenging to engage online, so the remote format will be eliminated, says Astrid Fingerhut. Groups will also be smaller, with more time allocated for demonstrations on the second day.
Since the hackathon, employees have access to enterprise accounts and have a better understanding of security settings; the AI environment and university policies have also evolved. Tom Kompare reports more detailed institutional guidelines on AI security. He anticipates being able to direct participants to specific rules rather than limiting himself to general prohibitions.
Allison Kallo believes that excessive assistance can foster apprehension, while encouragement helped participants explore with confidence. To enable other institutions to organize a comparable experience, the team has compiled resources outlining its method. Astrid Fingerhut invites them to retain useful tools and adapt them to their own environment.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.