(Written by Ry, a TTS-built custom Copilot AI Chatbot. Edited by Freedom Baird)

When a project required coding, Jake Dempsey used to have two choices: figure out who to ask—or give up on doing it himself. Now, with the help of generative AI, he’s building his own solutions, writing functional scripts, and solving problems independently—all without formal programming training.

 Dempsey, a manager in Tufts University’s Educational Technology Services, has been experimenting with tools like ChatGPT and Claude to bridge a gap that many non-programmers know well: the distance between an idea and a working solution. In the past, tasks that required even modest coding—like transforming data, automating repetitive processes, or integrating systems—often meant relying on a colleague with coding expertise. Today, Dempsey is increasingly able to tackle those challenges on his own. By clearly describing what he wants a tool to do, he collaborates with AI systems to generate scripts—often in Python—that he can test, refine, and put to use.

The results are practical and immediate. Dempsey’s examples range from relatively simple tasks, like merging CSV files or restructuring data, to more complex projects involving APIs and automation workflows. What these projects have in common is not technical elegance, but usefulness. As Dempsey describes it, the goal isn’t perfection—it’s“good enough” solutions that save time and reduce dependency on others. That mindset allows him to move quickly: generate code, test it on a small dataset, iterate, and improve. When something breaks, he feeds the error messages back into the AI and continues refining until it works. 

Crucially, Dempsey approaches this work with a strong sense of responsibility. His advice to others is to start small, avoid sensitive data, and stick to approved tools like Microsoft Copilot. He also underscores an important principle: the AI may generate the code, but the human remains accountable for how it’s used. That includes understanding limitations, questioning outputs, and recognizing when a project might be too risky or complex to pursue independently. For those looking for guidance, Tufts provides up-to-date resources on responsible AI use at the TTS Generative AI website. 

For Tufts Technology Services staff—and really anyone working in technology services in higher education—Dempsey’s work offers a compelling takeaway: you don’t need to be a trained developer to start solving technical problems. As staff and faculty become more comfortable building lightweight tools themselves, they are better positioned to support innovative classroom practices, streamline administrative tasks, and model a spirit of exploration for students.

At the same time, Dempsey’s approach highlights an important balance. AI can expand what individuals can do on their own—but it doesn’t replace the need for thoughtful decision-making, collaboration, or technical expertise when it matters most. In that sense, his work isn’t just about writing code without coding. It’s about developing a new kind of fluency: knowing how to collaborate with AI to get something useful done.

For anyone who has ever thought, “I wish I could just build this myself,” Dempsey’s experience is an encouraging place to start.